Skip to main content

Full text of "The Conference Proceedings of the 1999 Air Transport Research Group (ATRG) of the WCTR Society"

See other formats


THE UNO A VIA TION MONOGRAPH SERIES 

UNOAI Report 99-6 

The Conference Proceedings of the 1999 Air 
Transport Research Group (ATRG) of the 
WCTR Society 



Volume 2 



Editors 

Anming Zhang 
Brent D. Bowen 



September 1999 



UNO 

Aviation Institute 

University of Nebraska at Omaha 

Omaha, NE 68182-0508 



1999, Aviation Institute, University of Nebraska at Omaha 



UNO Aviation Institute Monograph Series 

Michaela M. Schaaf, Series Editor 
Mary M. Schaffart, Production Assistant 



Host Organization 

The University of Nebraska at Omaha, Dr. Nancy Belck, Chancellor 

Vice Chancellor for Academic Affairs, Dr. Derek Hodgson 

College of Public Affairs and Community Service, Dr. David Hinton, Dean 

Department of Public Administration, Dr. B. J. Reed, Chair 

Aviation Institute, Dr. Brent D. Bowen, Director 



Funding Support 

NASA National Space Grant College and Fellowship Program & NASA EPSCoR, 

Dr. Julius Dasch, Program Manager 
NASA Nebraska Space Grant & EPSCoR Programs, Dr. Brent D. Bowen, Director 



Publication 

The UNO Aviation Institute Monograph Series is published at the University of Nebraska 
at Omaha, 6001 Dodge Street, Omaha, NE 68182. 

Published as a not-for-profit service of the Aviation Institute. Funded in part by a grant 
from the NASA National Space Grant College and Fellowship Program. 

The University of Nebraska does not discriminate in its academic, employment or 
admission policies and abides by all federal, state, and regental regulations pertaining to 
same. 



The University of Nebraska at Omaha 
Aviation Institute 
Monograph Series 



Mission 



The UNO Aviation Institute Monograph Series began in 1994 as a key component of the 
education outreach and information transfer missions of the Aviation Institute and the 
NASA Nebraska Space Grant & EPSCoR Programs. The series is an outlet for aviation 
materials to be indexed and disseminated through an efficient medium. Publications are 
welcome in all aspects of aviation. Publication formats may include, but are not limited 
to, conference proceedings, bibliographies, research reports, manuals, technical reports, 
and other documents that should be archived and indexed for future reference by the 
aviation and world wide communities. 



Submissions 



Aviation industry practitioners, educators, researchers, and others are invited to submit 
documents for review and possible publication in the monograph series. The required 
information is listed in the Submission Checklist, found on the world wide web at: 

http://cid.unomaha.edu/~nasa 
Select UNOAI Monograph Series, select Submission Checklist. 



Dissemination 



The UNO Aviation Institute Monograph Series is indexed in various databases such as 
Educational Research Information Clearinghouse (ERIC), Transportation Research 
Information Services (TRIS), Aviation TradeScan, NASA Scientific & Technical Reports 
(STAR), and the Library of Congress. The series is also cataloged in the UNO Library, 
which is a member of the Online Computer Library Center (OCLC), an international 
bibliographic utility. OCLC's Union Catalog is accessible world wide and is used by 
researchers via electronic database services EPIC and FirstSearch and is also used for 
interlibrary loans. In addition, copies have been provided to the University of Nebraska - 
Lincoln and the University of Nebraska at Kearney Libraries. Copies are also provided to 
the Nebraska Library Commission, the official archive of state publications. 



Ordering 



UNO Aviation Institute monographs are available from the UNO Aviation Institute, 
Allwine Hall 422, 6001 Dodge Street, Omaha, NE 68182-0508. Order information is also 
available on the world wide web at <http://cid.unomaha.edu/~nasa>, select UNOAI 
Monograph Series. 



University of Nebraska at Omaha Aviation Institute 
Aviation Monograph Series 

Recent monographs in the series include: 

99-5 thru 99-8 The Conference Proceedings of the 1999 Air Transport Research Group of the WCTR Society 

99-4 Selected Papers on Aviation Security 

99-3 The Airline Quality Rating 1999 

99-2 NASA EPSCoR Nebraska Preparation Grant: Year 1 

99-1 NASA Nebraska Space Grant Consortium 1995-1999 Self-Evaluation 

98-6 thru 98-9 The Conference Proceedings of the 1998 Air Transport Research Group (ATRG) of the WCTR Society 

98-3 thru 98-5 The Symposium Proceedings of the 1998 Air Transport Research Group 

98-2 Aviation Security: Responses to the Gore Commission 

98-1 The Airline Quality Rating 1998 

97-9 The Airline Quality Rating 1997 

97-2 thru 97-8 The Conference Proceedings of the 1997 Air Transport Research Group (ATRG) of the WCTR Society 

97-1 Aviation Institute Self Study Report for the Council on Aviation Accreditation 

96-4 The Airline Quality Rating 1996 

96-3 NASA and Ethics: An Annotated Bibliography 

96-2 The Image of Airport Security: An Annotated Bibliography 

96-1 Concentration and Contestabiiity in the Deregulated United States Airline Industry 

95-2 The Nebraska Initiative for Aerospace Research and Industrial Development 

A complete listing of monographs is available at http://cid.unomaha.edu/~nasa ; select UNO Aviation Monograph Series. 

To Obtain Monographs 

Complete this form and include a check or purchase order made payable to the Aviation Institute. Orders within the U.S. are 
$7.50 (U.S.) per monograph, and international orders are $10.00 (U.S.) to cover the costs of printing, shipping, and handling. 
Allow 4-6 weeks for delivery. Please forward this request to: Aviation Institute, University of Nebraska at Omaha, 6001 Dodge 
Street, Omaha, NE 68182-0406. Phone: 402-554-3424 or 1-800-3 FLY UNO; Fax: 402-554-3781; E-mail: nasa@unomaha.edu 

Name 



Company 
Address 



City, St., Zip 

Country 

Phone 



E-mail 



Quantity 


Monograph # 


Unit Cost 


Total Cost 








$ 








$ 








$ 


TOTAL ENCLOSED 


$ 



This series is co-sponsored by the NASA Nebraska Space Grant Consortium 



ATRG Networking Committee 



Dr. Tae H. Oum (President) 

University of British Columbia 
Vancouver, BC, Canada 

Prof. John Black 

University of New South Wales 
Sydney, NSW, Australia 

Prof. John Brander 

University of New Brunswick 
Fredericton, N.B., Canada 

Prof. Kenneth Button 

George Mason University 
Fairfax, VA, U.S.A. 

Prof. Jaap DeWit 

Dept. of Civil Aviation 
The Hague, Netherlands 

Dr. Christopher Findlay 

University of Adelaide 
Adelaide, SA, Australia 

Prof. Mark Hansen 

University of California at Berkeley 
Berkeley, CA, U.S.A. 

Dr. Keith J. Mason 

Cranfield University 
Cranfield, Bedford, U.K. 

Prof. Hirotaka Yamauchi 

Hitotsubashi University 
Tokyo, Japan 



Prof. Joseph Yossi Berechman 

Tel Aviv University 
Ramat Aviv, Israel 

Dr. Brent Bowen 

University of Nebraska at Omaha 
Omaha, NE, U.S.A. 

Prof. Jean Bresson 

Ecole Natinale De L' Aviation Civile 
Toulouse, France 

Prof. Anthony Chin 

National University of Singapore 
Kent Ridge, Singapore 

Prof. Martin Dresner 

University of Maryland 
College Park, Maryland, U.S.A. 

Prof. David W. Gillen 

Wilfred Laurier University 
Waterloo, Ontario, Canada 

Dr. Paul Hooper 

National University of Singapore 
Singapore 

Dr. Bill Swan 

Boeing Commercial Airplane Group 
Seattle, WA, U.S.A. 



About the Editors 



Dr. Anming Zhang, the Acting Head of the Department of Economics and Finance at 
City University of Hong Kong, joined the City as an associate professor in 1996 after 
teaching at University of Victoria, Canada, for six years. He received a BSc from 
Shanghai Jiao Tong University, MSc and PhD (1990, Economics and Business Admin.) 
from University of British Columbia. Dr. Zhang has published more than 20 research 
papers in the areas of industrial organization, international trade, and transportation. He 
received the Yokohama Special Prize for an Outstanding Young Researcher, awarded at 
the 7"" World Conference on Transportation Research (WCTR), Sydney, Australia, July 
1995. He won again the Overall Best Paper award (with Tae Oum and Yimin Zhang) at 
the 8"" WCTR, Antwerp, Belgium, July 1998. Dr. Zhang has also done extensive 
consultancy work for government and industry. 

Dr. Brent D. Bowen is Director and Professor, Aviation Institute, University of 
Nebraska at Omaha. He has been appointed as a Graduate Faculty of the University of 
Nebraska System-wide Graduate College. Bowen attained his Doctorate in Higher 
Education and Aviation from Oklahoma State University and a Master of Business 
Administration degree from Oklahoma City University. His Federal Aviation 
Administration certifications include Airline Transport Pilot, Certified Flight Instructor. 
Advanced-Instrument Ground Instructor, Aviation Safety Counselor, and Aerospace 
Education Counselor. Dr. Bowen' s research interests focus on aviation applications of 
public productivity enhancement and marketing in the areas of service quality evaluation, 
forecasting, and student recruitment in collegiate aviation programs. He is also well 
published in areas related to effective teaching. His professional affiliations include the 
University Aviation Association, Council on Aviation Accreditation, World Aerospace 
Education Organization, International Air Transportation Research Group, Aerospace 
Education Association, Alpha Eta Rho International Aviation Fraternity, and the 
Nebraska Academy of Sciences. He also serves as program director and principal 
investigator of the National Aeronautics and Space Administration funded Nebraska 
Space Grant and EPSCoR Programs. 



A TRG President's Foreword 

The Air Transport Research Group of the WCTR Society was formally launched as a special 
interest group at the T"" Triennial WCTR in Sydney, Australia in 1995. Since then, our membership base 
has expanded rapidly, and includes nearly 600 active transportation researchers, policy-makers, industry 
executives, major corporations and research institutes from 28 countries. Our broad base of membership 
and their strong enthusiasm have pushed the group forward, to continuously initiate new events and 
projects which will benefit aviation industry and research communities worldwide. 

It became a tradition that the ATRG holds an international conference at least once per year . As 
you know, the 1997 conference was held in Vancouver, Canada. Over 90 papers, panel discussions and 
invited speeches were presented. In 1998, the ATRG organized a consecutive stream of 14 aviation 
sessions at the S"' Triennial WCTR Conference (July 12-17: Antwerp). Again, on 19-21 July, 1998, the 
ATRG Symposium was organized and executed every successfully by Dr. Aisling Reynolds-Feighan of 
the University College of Dublin. 

In 1999, the City University of Hong Kong has hosted the 3"^ Annual ATRG Conference. Despite 
the delay in starting our conference sessions because of Typhoon Maggie, we were able to complete the 
two-day conference sessions and presentation of all of the papers. On behalf of the ATRG membership, I 
would like to thank Dr. Anming Zhang who organized the conference and his associates and assistants for 
their effort which were essential for the success of the conference. Our special thanks go to Professor 
Richard Ho, Dean of the School of Business and Economics of the University for the generous support 
for the conference. Many of us also enjoyed the technical visit to the new Hong Kong International 
Airport (Chep Lok Kok). 

As you know, Professor Jaap de Wit and I look forward to welcoming you to University of 
Amsterdam on July 2-4, 2000 for the 4"' Annual ATRG Conference. 

As in the past, the Aviation Institute of the University of Nebraska at Omaha (Dr. Brent Bowen , 
Director of the Institute) has kindly agreed to publish the Proceedings of the 1999 ATRG Hong Kong 
Conference (being co-edited by Dr. Anming Zhang and Professor Brent Bowen). On behalf of the ATRG 
members, I would like to express ray sincere appreciation to Professor Brent Bowen, Mary M. Schaffart 
and the staff of the Aviation Institute of University of Nebraska at Omaha for the effort to publish these 
ATRG proceedings. Also, I would like to thank and congratulate all authors of the papers for their fine 
contribution to the conferences and the Proceedings. Our special thanks are extended to Boeing 
Commercial Aviation - Marketing Group for the partial support for publication of this proceedings. 

Finally, I would like to draw your attention to the ATRG newsletter and the ATRG website 
(www.commerce.ubc.ca/atrg/ ) which will keep you informed of the ATRG operations and forthcoming 
events. On behalf of the ATRG Networking Committee, I would appreciate it very much if you could 
suggest others to sign up the ATRG membership. Thank you for your attention. 

Tae H. Oum 
President, ATRG 

ATRG c/ o Prof. Tae H. Oum 

Faculty of Commerce and Business Administration, 

University of British Columbia, 2053 Main Mall 

Vancouver, B.C., V6T lZ2Canada 

E-mail: Atrg(5commerce.ubc.ca 



CONTACT INFORMATION 



The Conference Proceedings of the 1999 Air Transport Research Group of the WCTR Society 
International Conference on Air Transportation Operations and Policy 

June 6-8, 1999: City University of Hong Kong, Hong Kong 

Dr. Anming Zhang 

Coordinator, Transportation Operations and Policy (TOP) Group 

Center for Competitiveness 

Acting Head, Department of Economics and Finance 

Faculty of Business 

City University of Hong Kong 

Phone: +852-2788-7342 

Fax: +852-2788-8806 

E-mail: EFANMING@cityu.edu.hk 



Air Transport Research Group (ATRG) 
International Conference on Air Transportation Operations and Policy 

City University of Hong Kong 
June 6-8, 1999 



I'he Conference 



The ATRG held its 3'" Annual Conference at the City 
University of Hong Kong Campus in June 1999. 

The 1999 Conference contained 13 aviation and 
airport sessions. Over 40 research presentations were 
featured on topics pertaining to airports and aviation; 
these titles are listed on the ATRG website 
(http://www.commerce.ubc.ca/alrg/). 



The Proceediniis 



Once again, on behalf of the Air Transport Research 
Group, the University of Nebraska at Omaha Aviation 
Institute has agreed to publish the Proceedings of the 
ATRG Conference in a four-volume monograph set. 



Volume 1 



Session 1: Infrastructure Financing 

E. CHU, Implementation of Satellite-based Systems 

for Civil Aviation in the 2P' Century. 
P. LOK, Prospects of Air Transport Industry in China. 

Session 2A: Strategic Alliance 

K. TAKADA, T. YAI, & M. HARADA, Evaluation of 
Airline Alliance in International Aviation 
Market Using Equilibrium Analysis. 

B. KLEYMANN. Future Developments in the 
Structure of Airline Alliance Networks. 

M. LI, Distinct Features of Lasting and Non-la.sting 
Airline Alliances. 

J. BRUECKNER & Y. ZHANG, Scheduling 

Decisions in an Airline Network: A Hub-and- 
Spoke System's Effect on Flight Frequency, 
Fares and Welfare. 



Proceedings Order Information 



The Proceedings of the 1999 ATRG Conference are 
contained in a four-volume monograph set. Orders 
within the U.S. are $7.50 (U.S.) per monograph 
volume, and international orders are $10.00 (U.S.) per 
monograph volume to cover the costs of printing, 
shipping, and handling. Allow 4-6 weeks for delivery. 



Please forward requests to: 

UNO Aviation Institute 
6001 Dodge Street 
Allwine Hall 422 
Omaha, NE 68182-0406 

Phone: 402-554-3424 or 1-800-3 FLY UNO 

Fax: 402-554-3781 

e-mail: nasa@unomaha.edu 

http://cid.unomaha.edu/~nasa 



Session 2B: Airports and the 
Environment 

p. MORRELL & H. LU, Noise Social Cost and 

Commercial Flight Charge Mechanism — A 
Case Study of Amsterdam Schiphol Airport. 

H. JATMIKA & J. BLACK, Economic Impacts and 
Land-use Change around Major Airports: 
Some Examples from Sydney Kingsford 
Smith Airport Based on an Eighty-year 
Analysis. 



Volume 2 



Session 3A: Liberalization 

D. GILLEN, R. HARRIS, & T. OUM, A Model for 

Measuring Economic Effects of Bilateral Air 
Transport Liberalization 

J. DeWIT, p. UITTENBOGAART, & T. WEI-YUN, 
Hubbing and Hub-bypassing: Network 
Developments in a Deregulated European 
Airport System 

B. GRAHAM, Environmental Sustainability, Airport 
Capacity and European Air Transport 
Liberalization: Irreconcilable Goals? 

Y. CHANG & G. WILLIAMS, Civil Aviation 
Development in the Taiwan Area. 



Volume 2 (continued) 



Session 3B: Airport Economics and 
Management 

B. MANDEL, Airport Choice and Competition - A 

Strategic Approach. 
K. YOO & Y. LEE, A Study on the Flight Service 

Network for Incheon International Airport to 

be a Successful Hub Airport in Northeast 

Asia. 
P. ACHARJEE & K. LUMSDEN, Airfreight from a 

Process Concept. 

Session 4A: Airline Regulation and 
Privatization 

V. TAMMS, Re-examining the Slot Allocation 

Problem. 
H. SERISTO: Regulation as a Driver for International 

Airline Alliances. 
H. YAMAUCHI, Air Transport Policy in Japan: 

Policy Change and Market Competition. 

Session 4B: Airline-Airport Interaction 

M. HANSEN, D. GILLEN, & R. DJAFARIAN- 
TEHRANI, Aviation Infrastructure 
Performance and Airline Cost: A Statistical 
Cost Estimation Approach. 

E. PELS, P. NIJKAMP, & P. RIETVELD, Relative 
Efficiency of European Airports. 

G. NERO & J. BLACK, A Critical Examination of an 
Airport Noise Mitigation Scheme and an 
Aircraft Noise Charge: The Case of Capacity 
Expansion and Externalities at Sydney 
(Kingsford Smith) Airport. 



Volume 3 



Session 5A: Financial Performance 

R. GRITTA, E. FREED, G. CHOW, The Effects of 
Operating and Financial Leverage on the 
Stability of Airline Returns Over Time: The 
Contrast between Southwest, Delta and 
USAir. 

S. GUDMUNDSSON, A Factor Analytical Study of 
Airline Management: The Case of New 
Entrant Airlines. 

J. PARK, A. ZHANG, & N. PARK, Strategic Alliance 
and Firm Value: A Longitudinal Study of the 
British AirwaysAJSAir Alliance. 



Session 5B: Service Quality 

R. KADUCK, Intransit Preclearance at Canadian 

Airports. 
D. RHOADES & B. WAGUESPACK, Jr., Judging a 

Book by its Cover: The Relationship between 

Service and Safety Quality in U.S. National 

and Regional Airlines. 
A. STEPUSHIN, Statistical Approach for Evaluation 

of Airport Pavement Functional Life. 

Session 6A: Cost Competitiveness 

K. MASON, The Propriety of Business Travellers to 

Use Low Cost, Short Haul Airlines. 
M. DRESNER, J. LIN, & R. WINDLE. Determinants 

of Price Reactions to Entry in the US Airline 

Industry. 
J. BRESSON, Yield Management Simulation: 

Appraisal of Airline Revenue in Different 

Cases. 
F. ALAMDARI, Airline In-flight Entertainment: The 

Passengers' Perspective. 



Volume 4 



Session 6B: Demand for Air Traffic 

K. BUTTON & R. STOUGH, Technology Intensive 
Employment and Hub Airports. 

A. ROMAN, Changes in Flight Frequency and Point- 
to-Point Service: Effects on Aviation 
Demand. 

S. BRATHEN, K. ERIKSEN, H. HJELLE, & M. 

KILLI, Methods for Economic Appraisal in 
the Norwegian Aviation Sector. 

T. ISHIKURA & H. INAMURA, A Study on Inter- 
city Air/Rail Traveler's Behavior 
Considering Arrival Margin Time. 

Session 7A: Airport Regulation and 
Privatization 

P. HOOPER, R. CAIN, & S. WHITE, The 

Privatisation of Australia's Airports. 
P. FORSYTH, Regulating Access to Airport Facilities. 

Session 7B: Asian Aviation Markets 

A. CHIN, P. HOOPER, & T. OUM, The Impacts of 
the Asian Economics Crises on Asian 
Airlines: Short-run Responses and Long-run 
Effects. 

M. Zl, Asia-Pacific Airlines Amidst the Asian 
Economic Crisis. 

H. CHEN, The Reform and Performance of Air 
Transportation Industry of China. 

X. XIE, An Analysis of the Development and Targets 
of China Civil Aviation Industry. 

Z. SHON, Y. CHANG, & C. LIN, Direct Flights 

Across Taiwan Strait and Their Impacts on 
Eastern Asian Air Transportation Market. 



A MODEL FOR MEASURING ECONOMIC EFFECTS OF 
BILATERAL AIR TRANSPORT LIBERALIZATION 



by 



David Gillen*, Richard Harris** and Tae Hoon Oum*** 



The earlier version of this paper was presented at the International Colloquium on Air 
Transportation, Toulouse, France, November 17-19, 1998, and at the American 
Economics Association (AEA) Conference (January 3-6, 1999: New York). The authors 
are grateful to Michael O'Connor, Steven Lewis- Workman and Jonathan Harvey for their 
work on simulation programs, and also to Transport Canada for funding this research 
through a contract to HLB Inc. 



* University of California & Wilfrid Laurier University 

** Simon Fraser University 

*** University of British Columbia 



dgillen@newton.berkeley.edu 

rharris@sfu.ca 

tae.oum@commerce.ubc.ca 



A MODEL FOR MEASURING ECONOMIC EFFECTS OF 
BILATERAL AIR TRANSPORT LIBERALIZATION 

David Gillen, Richard Harris, and Tae Hoon Oum 

ABSTRACT 
In this paper, we develop a model with which allows us to measure not only the changes 
in equilibrium outcomes and welfare consequences of liberalizing a bilateral air transport 
agreement, but also the distribution of the gains and losses to carriers and consumers of each 
bilateral country and those of the third foreign countries. Our model also allows to measure the 
effects of changes in a bilateral agreement on the amount of traffic diversion between the direct 
bilateral routes and the indirect routes via a third country. We also provide an extension of our 
model to a case of oligopoly market outcome (Coumot Nash equilibrium). In our model, quality 
aspects are treated in the framework of hedonic price theory by specifying the quality-adjusted 
price (quantity) as a multiplication of the observed price (quantity) by the reciprocal quality 
index function (the quality index function). 

Numerical simulations were conducted to measure the effects of changing the following 
major policy levers in a bilateral air transport agreement: 

• Removing price regulation while retaining frequency and entry restrictions 

• Removing price and entry regulation while retaining frequency restrictions 

• Removing frequency regulations while retaining price and entry regulations 

• Removing frequency and entry regulations while retaining price regulation 

• Removing price and frequency regulations while retaining entry restriction 

• Removing all price, frequency and entry regulations (de facto, open skies) 

The application to the case of the Canada- Japan bilateral agreement show the following 
results: 

• Frequency competition without freeing entry or price regulation neither increase airline 
profits nor improve consumer welfare. Frequency competition with entry freedom increases 
the welfare of the nation whose carrier enters the market, i.e., the nation with lower cost 
carriers. 

• Pricing freedom with frequency regulation increases the welfare of the nation with a larger 
share of passengers on the bilateral markets more than other countries. The benefits of price 
competition becomes more than doubled if entry is also freed. 

• Overall, allowing entry of new carriers increase the overall welfare the most, followed by the 
price freedom. Just the removal of frequency restrictions has the least effect on consumer 
welfare. 

• The complete liberalization of pricing, frequency and entry leads to the welfare maximizing 
market outcome. Oligopoly solution (Coumot Nash equilibrium) increase carrier profits 
while reducing consumer surplus substantially. 

• The effects of liberalization of price and frequency regulations on 0-D traffic volume, 
carrier profits and consumer surpluses are greater when the model takes into account of the 
third country routing possibilities. 



A MODEL FOR MEASURING ECONOMIC EFFECTS OF 
BILATERAL AIR TRANSPORT LIBERALIZATION 



1. Introduction 

Although telecommunications, financial and maritime services have been incorporated in 
the General Agreement on Trade in Services (GATS) being governed by the World Trade 
Organization (WTO), most of the commercial air transport issues are not incorporated in 
the GATS. For more than 50 years, the issues involving commercial rights on 
international air transport have been governed by the bilateral agreements between each 
pair of countries involved. The bilateral framework for negotiating commercial aviation 
rights was adopted at the 1944 Chicago Convention. The Bermuda Agreement, the first 
bilateral air treaty signed between the United States and the U.K. in 1946 (Bermuda I) has 
served as the legal fi-amework for the fiiture bilateral agreements to follow. Bermuda I 
introduced capacity regulation, designation of carriers and routes, and system 
requirements for fair and equal opportunity for carriers involved. This complex system of 
regulation of commercial air transport has come under increasing criticism and pressure 
during the last two decades. 

Soon after the deregulation of their domestic air transport markets in 1978, the U.S. 
government turned their attention to liberalize the bilateral agreements with foreign 
countries'. By virtue of the sheer size of their domestic market and the strength of its 
carriers, the U.S. was able to impart a pro-competitive approach on a large number of 
nations. The U.S. government used the new liberal bilaterals as a means of putting 
pressure on some reluctant governments. The UK and Germany were pressured by 
expansion of air service between the U.S. and Belgium and The Netherlands. A new 
liberal agreement with South Korea put pressure on Japan. Also, the U.S. took advantage 
of 5th fi-eedom rights in certain countries to circumvent restrictions in neighboring 
nations with more restrictive bilaterals. This liberal bilateral approach was successful. 
The "Open Skies" campaign which the U.S. government began in 1992 was stimulated by 
strong criticism of the restrictive terms of Bermuda II." Although these U.S. initiatives 
and the general movements towards fireer goods and services trade have been successful 
for liberalizing air transport system to and from the U.S., the system of bilateral 
agreements between countries remain entrenched in the intemational air transport system. 



' For a nearly comprehensive measurement of the effects of the U.S. domestic deregulation on air carriers 
and travelers, please refer to Morrison and Winston (19S6). 

" In 1977, the UK renounced the Bermuda Agreement. The Bermuda II .Agreement \\as accepted by the 
US and the UK and was aimed at restructuring the air relationships that had developed after 1945. This 
agreement was, in many ways more restrictive than the agreement that preceded it and was never a model 
for US bilateral air transport agreements. 



For the next decade or two, liberalization of international air transport markets will 
depend nearly entirely on bilateral negotiations between countries involved. 

In any bilateral negotiation, including general trade negotiations, parties to a bilateral 
negotiations on air transport are concerned very much with who gains, who loses and 
what impact the proposed change in the bilateral might have on traffic diversion to third 
country routings. Both of the countries involved need to better understand the 
consequences on the consumers and carriers of each bilateral partner and of any third 
party foreign nations of the proposed changes to the rules and restrictions governing the 
air transport bilateral. Although bilateral negotiators would like to take account of these 
complex effects in making decisions, air transport researchers have not directed sufficient 
effort to develop models which allow one to measure these complex consequences of the 
bilateral air transport liberalization. 

Two previous studies have developed measures of the welfare gains from international air 
transport liberalization. Street, Smith and Savage (1994) measure both the gains in cost 
efficiency and improvements in service quality that would be caused by introducing 
greater competition in the Australia's international air transport markets. This paper is 
close to the current paper in that it attempted to measure gains and losses from bilateral 
liberalization to the Australian carriers and consumers. They use conventional "triangle" 
analysis via perfect competition model while doing some ad hoc adjustments for quality 
of services including flight frequency. The equilibrium outcomes are computed without 
full iteration of the interactive nature of demand and supply functions. They show that 
liberalization does not always yield a net welfare gain to Australian economy although, 
by definition, it will increase the world's welfare. 

In another study, Findlay, Hufbauer and Jaggi (1996) measured the potential cost savings 
to users from the Asia-Pacific regional open skies regime. They assumed that the gains 
are all in the form of airlines' cost savings, all of which are transferred to consumers 
through competition. On the basis of this coarse calculation they show that the cost 
savings to users of seven countries (Australia, Canada, Hong Kong, Japan, Korea, 
Malaysia and Singapore) alone is S2 1 .7 billion in 2010. 

The purpose of this paper is to develop a model with which to assess the effects of 
changes in a bilateral air transport agreement governing the supply of air transport 
services on the distribution of benefits and costs to bilateral partner nations' carriers, 
consumers and foreign carriers and consumers. Our approach differs from the traditional 
benefit-cost analysis of policy alternatives. Our focus is in the distribution of benefits and 
costs to various stakeholders: consumers and carriers of each bilateral partners and of 
third countries ' while cost-benefit methodology focus on measuring aggregate economic 
benefits and costs for a given nation or region. Since trade policy researchers have 
concentrated on the distribution of the impacts of change in trading mles across nations 
and on identifying the winners and losers, methodologically we borrow significantly from 



' Governments are also a major gainer or loser through changes in net ta.\/subsidy revenues particularly as 
indirect taxation is an important feature or the tax framework or industrial or subsidy policies are in place. 



the literature on trade policy analysis. In particular, our main model is the monopolistic 
competition model adopted from Spence (1976), Dixit and Stiglitz (1977), Krugman 
(1980), and Harris (1984). We also extend our model to obtain the effects of a different 
market structure (i.e., oligopoly) on the distribution of the gains and losses of bilateral air 
transport liberalization. 



2. Key Policy Levers in Bilateral Air Transport Liberalization 

The key commercial rights to be negotiated in a bilateral air agreement are pricing, 
capacity (new entry, frequency and aircraft size), and carrier and route designations. 
Carrier and route designation form a barrier to competition in several ways. First, 
bilateral air treaties normally limit the number of carriers who can serve the bilateral 
markets. For example, most bilateral agreements allow one carrier from each country to 
serve markets between the two countries: e.g., Korean Air and Air Canada in the Canada- 
Korea bilateral markets. Second, the cities and/or airports a designated foreign carrier 
can serve are normally specified. For example. Air Canada can serve Kansai 
International Airport only from Vancouver or Toronto while Canadian is allowed to serve 
only Tokyo's Narita Airport and Nagoya. Third, most bilaterals do not allow the fifth 
freedom rights (beyond right) to foreign carriers.'* Clearly, removal or relaxation of the 
carrier/route designation clause is likely to induce competitive entry by new carriers as 
well as encourage entry into new routes and/or airports by the existing carriers. 

Pricing regulations in bilateral agreements usually take one of the following forms. First, 
all carriers may be required to use the lATA set fares. Second, when only one carrier 
from each country serves the market the bilateral agreement may require the two carriers 
to agree on a uniform price. The third option is the so-called "single-disapproval" 
pricing regime that allows for one of the two governments to disapprove a carrier's fare 
proposal. In this case, a carrier's proposed fares are usually disapproved by the foreign 
government. The fourth option is the "double disapproval" regime. Under this regime, 
both governments are required to agree in order to disapprove a carrier's proposed fares. 
Airiines will have nearly complete pricing freedom under the double disapproval regime. 
Naturally, removal or relaxation of the pricing regulation increases competition. 

The seat capacity that the carriers of a country can offer in aggregate are usually restricted 
in bilateral agreements.^ Although many bilateral agreements allow the tradeoff between 
frequency of services and aircraft size used, in many bilateral routes the cost 



■" Fifth Freedom rights must be negotiated separately and must include the third count>'. 
' The nature of the capacity controls to either carrier in a bilateral is important. RestTicti%e regimes require 
agreement bet\veen designated carriers and approval by both authorities. Canada has very restrictive 
capacity regimes in all areas except with the United Kingdom, Germany and the Netherlands. In its largest 
twelve international markets, Canadian government has the right to designate more than one carrier in 
eleven of them. This multiple designation benefit could be negated with overall cap on capacity in the case 
of Hong Kong, Japan and Ausn-aha, while in France and Italy there would need to be an agreement 
between the incumbent carriers. Canada has the right to designate additional carriers, without capacity 
restrictions for 3rd and 4th freedom ser(.ices to the Linited Kingdom. Germany. Netherlands, Jamaica, 
.Mexico and Trinidad. 



characteristics and the need to offer near daily flights limit the aircraft choice practically 
to only one or two aircraft types. Therefore, for simplicity of our analysis we will 
analyze the effects of allowing frequency competition only. We choose the optimal 
aircraft type for each route based on the unit cost per passenger. 

Although other factors including access to airport slots and facilities could influence 
competitive outcomes significantly in bilateral air transport markets, in this study we will 
examine only the effects of removing or relaxing the pricing, capacity and entry 
regulations on consumers and carriers of each country and for each nation. 



3. Model Development and Estimation 

Our model uses a two-stage approach. In the first stage, the analytic foundations are laid by 
specifying demand and cost fianctions, market clearing conditions for demand and supply, 
and the equilibrium quantities and prices. In the second stage, numerical simulation is used 
to fmd each carrier's equilibrium traffic volume, price, and frequency. 

Since air transport services are supplied in a network, the model must integrate all of the 
routing possibilities for each Origin-Destination (OD) market for all OD markets in the 
network. The starting point is to identify all of the OD market in the region, and then, 
map out the alternative routings for each of the OD pairs. In Figure 1, we illustrate a 
simplified version of such a network in the North Pacific. It is simplified to include only 
the OD traffic between Vancouver (YVR), Tokyo (NRT), Seattle (SEA) and Seoul (SEL). 
In this international network there are four countries, each country with a single airport, 
and three (3) potential OD markets in each country. For example, in Vancouver all 
originating traffic is assumed to be Vancouver based and destined for one of the three 
alternative destinations. Several alternative routes with several carriers serving a 
particular route can serve each OD market. Each route has one or more segments where a 
segment is a flight between two airports. Each carrier will carry passengers traveling on 
YVR-NRT, as well as passengers to YVR from other cities in Canada (Toronto-YYZ for 
example) and passengers traveling beyond NRT (to Seoul, for example). This means that 
passenger volume data must be captured separately by trip purpose, nationality, 
destination, fare class, route and carrier used. 

An important question is what will happen to the characteristics of this network in 
response to a bilateral liberalization beKveen Canada and Japan. While this liberalization 
will occur only to Japanese and Canada carriers and the network segments connecting 
Vancouver (YVR) and Tokyo (Narita or NRT), the evaluation of these effects will 
depend on the network wide demand and supply responses. For example, what portion of 
the traffic between Seoul (SEL) and YVR (or SEA and NRT) might be diverted through 
either NRT or YVR in response to the liberalization. Both the demand and supply 
specifications attempt to take into account these system wide interaction effects. 



Figure 1 



Illustrating Four Country North Pacific Air Transportation Network and Canada- Japan 
Liberalization 



KAL,AC 




KAL,AA,UA,NWT 



I I - Nodes 
► Segments 

-!^^ Carriers Serving Segments ' ^rea of Liberalization 



3.1 Characterization of Demand Mode! 

The demand for air travel depends upon fares, frequencies, other service attributes, and 
choice of carriers. Current bilaterals place restrictions on some or all of these variables. 
As bilaterals are liberalized, consumers will be given increased choices of new and 
different carriers. The demand model must be capable of incorporating differentiated 
products to correctly capture changes in the level and quality of air services delivered by 
incumbent as well as new entrant carriers. One way of introducing preferences for 
differentiated services and products is to treat each of them as a different variety of 
services (imperfect substitutes). This stems from the idea that there are more products 
than characteristics, and re-bundling characteristics results in another product. Car, stereo, 
vacation, restaurant, wine and air trips are well-defined products. However, each of these 
products can and does have many varieties; American, Australian, Canadian and German 
wines, for example. The way in which one can handle this 'many varieties' for a product 



can be handled by placing structure on the underlying preference functions thereby 
yielding manageable demand functions. 

Armington (1969) introduced this idea of variety and nationally differentiated products 
into the trade literature. Our demand model follows this idea to derive demands for 
"differentiated" services (imperfect substitutes). This approach was chosen for a number 
of reasons. When a new carrier enters the market it represents a new 'variety' of an already 
traded service. This approach also allows for market size as well as carrier shares to 
change. In the air travel liberalization model, the services are alternative route-carrier 
combinations serving a given origin-destination pair. The basic model is amended to 
incorporate non-price (including quality) factors which can vary across route-carrier 
alternatives. This "Armington" specification of demand is chosen with the goal of 
measuring all of the benefits of changing the rules affecting the supply of international air 
travel services. We need to measure all of the changes to consumer benefits at the level of 
the route-carrier choice, aggregate flight segment, route aggregate, and 0-D market. 

The demand side distinguishes between aggregate OD demand, and demand for 
individual routes connecting an OD pair. For every OD pair there is a home demand 
aggregator at each end of the route (e.g., Canadian residents traveling to Japan) for each 
fare class. Each OD group can thus be thought of as an individual consumer with a 
[Marshallian] demand curve given by 

Q=f(P) Equation 1 

Q is an index of total passenger demand over the particular OD market, served by a 

number of route-carrier combinations indexed r=l, ,R. For simplicity, we refer to any 

given r simply as a route. P is the real price index for this route and is a function of the 
individual prices p, for the route-carrier combination r. 

Q is referred to as the real quantity index of demand. It can also be interpreted as being 
measured in "utile" of aggregate real air service. Let q, be the individual route demand 
measured in conventional passenger unit terms and let p, be the route price. If we assume 
that the utility fiinction generating Q is positive linear homogenous so that: 

Q=U(g„...,qK) Equation 2 

Then there exists an exact price index function P(pi, .pj, which is dual to U(.). It is 

convenient to work with the price index fiinction rather than the utility or quantity index 
function. 

Adopting the "Armington" assumption, in our model the route-carrier combinations 
(different variety of air services serving an OD pair) are regarded as imperfect substitutes 
as reflected in the following CES price index function P(«): 



P{P\^-,Pr) = 



Y^s^p-; 



Equation 3 



Here the 4 are the weights on individual routes r and a is the AUen-Uzawa common 
elasticity of substitution between any two route pairs. Demand for route r given the level 
of aggregate Q, is given via Shephard's Lemma so that:' 

"^'^^Kp) ^ Equation 4 

In consumer equilibrium total consumer expenditure in this OD market is given by: 

E = Yp^q^ Equation 5 

r-l 

By construction E=PQ; actual expenditure on all routes is equal to the product of the 
price index and the aggregate real quantity index. 

In our multi-country demand analysis, welfare as well as market shares are dependent on 
quality differences. This would include the effect of national preferences on their flag 
carriers. In order to accommodate differences across routes in quality characteristics, the 
linear quality model is used (which is closely related to the hedonic price approach to 
quality adjustment).^ The quality index for route r with a vector of characteristics '.jc' 
could be represented by quality function as di^(f^t ^connections, national preference 
variable) where a^ is an increasing function in all variables corresponding to an increase 
in quality attributes such as frequency (/^). Our quality function, a^ (.), is specified as an 
iso-elastic function. 

The basic model starts with a definition of quality adjusted units of real service with the 
quality function a^-)- 

q' = a/xjq^ Equation 6 

The number q^ is referred to as the "unadjusted demand" or observed demand. In 
measurement terms it corresponds to the observable quantity of service r purchased by 
the consumer. In this case it corresponds to the number of passengers on a route in the 
relevant OD market. Higher a^ coefficients correspond to higher quality. Corresponding 
to the quality adjusted demand 17' there is a price /i"; i.e., the price per unit of 17'. 



' Given a list of prices p, on all routes, demand on route r is calculated in the following manner: compute 
the value of the aggregate price index P using equation 7; calculate the aggregate real quantity index Q 
using the aggregate demand curve (using equation 5); calculate individual route demands using equation 8 
' See Chapter 2, Tirole, Jean. Theory of Industrial Oreanization . for a complete exposition of this model. 



Given the linearity of q,* in q, it follows that: 

p'=(l/aXx))p,_ Equation 7 

The quality units are chosen such that for a service with quality level ^=l,p=p^. As the 
quality level rises for given /?„ the real price per unit of quality, p^' falls. Consumers 
actually buy the physical quantity q^ at price p„ but from a utiUty point of view purchase 
q* at price p* per unit of quality-adjusted demand. This is the figure that affects the 
calculation of consumer surplus, our measure of benefits. Depending on the nature of the 
supply side of the model, consumers take both/?^ and p' as given and choose q' and thus 

If we apply the same approach for deriving unadjusted route demand Sanctions to the case 
of quality-adjusted traffic volume, then the following expressions can be obtained. 

Q' =f(P') Equation 8 



P • = 



z^.p;- 



-I/O- 




Equation 9 



Equation 10 



To empirically implement such a procedure it is necessary to have information on the 
quality coefficients a^ so that the real quality adjusted prices p' can be calculated and 
substituted into the price index and demand functions. Having derived a quality-adjusted 
individual route demand via the same process outlined earlier, demand in observable 
units (passenger volumes) is given by: 



9.= 



S, 



r 



a^\P 



A- 1 Q' Equation 1 1 



One of the benefits of this particular demand specification is that it allows explicitly for 
calculating consumers' welfare and demand consequences of adding new routes or 
reducing route choices in a given OD market and for quality changes on those routes. In 
effect any changes in quality or in the number of carriers in the market (the variety effect) 
are represented in terms of changes in real prices. 

In equation 11 it is impossible to empirically distinguish between shifts in demand due to 
changes in 6^ and changes in a^. In order to identify the model we set all 5, equal to 1 . The 
implication of this assumption is that if all routes offer the same quality and prices, then 
by assumption, the demand would be equal on all routes. Demand differences therefore 
must be attributed in the benchmark and counterfactuals to either quality or price 



10 



differences across routes. As a simple illustration, suppose there is a market where there 
are only two routes, so R = 2 and both quality adjusted prices are equal to 1.0; thus 
demand on both routes is equal. Under these market conditions, the aggregate price index 
is given by: 

P'(2) = (Sl-'^ + Sl-'')-'" = {IS-^Y''" Equation 12 

Now suppose a new route is introduced that offers the same characteristics (price and 
non-price characteristics) as the previous routes so that its hedonic price is also unity, but 
now R = 3. In this case the new aggregate price index is given by: 

P'(S) = (^(ir" + t?(l)-" + ^(1)-")-"" = (S^T")-"" Equation 13 

We see in comparing these two equations that the real price index falls if a > 1; i.e. the 
elasticity of substitution between route/carriers exceeds unity. If cr = 2, P'(3)/P'(2) = 
0.81; an increase from 2 to 3 route-carrier combinations is equivalent to a 19 percent 
reduction in the real price of aggregate travel. When substituted into the demand flmction 
the increase in real quality adjusted demand is approximately rf • AP'. In the case of 
exit of a carrier from the market, the adjustment would be in the opposite direction. We 
can think of there being a taste for variety; more variety even if not every one consumes, 
makes people better off. In effect, variety will be valued in its own right, (Dixit and 
Stiglitz, 1977). 

This demand model is also explicitly structured to deal with the issue of inter-route 
substitution in a network context such as that outlined in the previous section. In Figiire 
1, for example, changes to the bilateral between Canada and Japan and liberalization of 
the of YVR-NRT market is likely to induce substitution between alternative routes 
connecting Japan and Canada, but passing through third countries. For example, a route 
such as SEA-NRT would be serviced by a U.S. carrier but with the liberalization the 
route SEA- YVR-NRT would draw some traffic. The extent of inter-route substitution in 
response to liberalization and its welfare consequences for consumers will depend on key 
parameters such as the carrier and route substitution elasticities and the relevant quality 
characteristics of competing routes. 



3. 2 Characterization of Supply Model 

Our supply model captures changes in the carrier costs in the market that could be 
influenced by restrictions imposed by the bilateral agreements. Airlines may respond to 
changes in those restrictions (e.g., flight frequency) by supplying more or less and/or new 
services. To the extent that economies of traffic density exists, the changes in service 
frequency and traffic volumes will change the unit cost 

In the demand model, any OD market may be served by a number of routes and each 
route is composed of one or more segments. In a nenvork of multiple CD's, it is probable 



that changes in one OD will affect passenger demands over the other OD's. For example, 
if higher frequencies are allowed between Canada and Japan not only will this OD market 
expand, but additional traffic may be garnered from passengers who are traveling to 
Korea via Japan. The flight segment YVR-NRT will carry more than simply the 0-D 
traffic and this will affect costs through economies of traffic density. The base unit for 
the cost function will therefore be the flight segment. Once the segment costs are 
calculated it is possible to construct a route cost by aggregating the relevant segment 
costs. The total carrier costs are calculated as the sum of all passenger costs and segment 
costs. 

In describing a carrier's costs we distinguish costs which vary by segment and those 
which vary by route. In many cases the source of cost differences will be in the airline 
system or station costs. For example, if carrier / were to extend its operation from point 
B to point C, when it was already in an AB market, the additional costs would include the 
increase in flight operating costs and passenger costs. However, since it is already 
serving airport B, the cost of adding an operation will be low. This is quite different from 
a case of entering an entirely new market. Clearly, both volume of passenger and flight 
frequency are important. 

Therefore, we define a carrier's total cost for a segment as follows: 

TC = vQ+ wF Equation 1 4 

where F is segment frequency and Q is segment passenger demand, v and w are unit cost 
per passenger and unit cost per flight, respectively. Carrier load factors are calculated as 
z = (Q/F)/G where G is seats per plane. Average per-passenger segment cost, u, can be 
computed by dividing the total segment cost by the number of passengers. The 
approximate average cost per passenger is obtained by dividing the total block hour costs 
(flight costs) for the segment plus the total passenger costs by the number of segment 
passengers. Therefore, the (total) unit cost per passenger on flight segment can be written 
as follows: 

u = TC/Q = wF/Q+v = (w/Gz)+v Equation 1 5 

The unit cost per passenger will change as the volume of passengers, flight frequency or 
load factors change. 



The per-flight operating cost per segment (w) was computed using the block-hour 
operating cost (for each aircraft type) for the U.S. carriers available using FORM 41 
data.' Since the block-hour costs on non-U.S. carriers were not available by aircraft type, 
the costs for American Airlines (AA) are adjusted for estimating the block-hour costs for 
foreign carriers. This involved taking account of the differential total factor productivity 



^ The cost data are taken from Aviation Daily, various issues. These figures are based on the information 
contained in the FORM 4 1 data series. 



12 



(TFP) and the aggregate input price index be^veen AA and the carrier under our 
consideration. Table 1 Usts the differential TFP and input price indices between the 
American Airlines and other carriers computed by Oum and Yu (1995). This information 
is used to estimate the block-hour costs for the carriers under consideration in this paper. 



Table 1: Productivity Index Table for Cost Calculations 



Productivity and Price Indices - 1993 




American = 1.00 






Numeric Code 


Airline 


TFP 


Price Index 


1 


American 


1.000 


1.000 


2 


United 


1.045 


1.035 


3 


Delta 


1.069 


1.075 


4 


Northwest 


1.124 


1.036 


5 


US Airways 


0.832 


1.034 


6 


Continental 


1.005 


0.891 


7 


Air Canada 


0.807 


0.881 


8 


Canadian 


0.860 


0.911 


9 


Japan Air Lines 


0.851 


1.421 


10 


All Nippon 


0.777 


1.432 


11 


Singapore Airlines 


0.958 


0.813 


12 


Korean Air 


0.988 


0.781 


13 


Cathay Pacific 


0.969 


0.926 


14 


Qantas 


0.875 


0.897 


15 


Thai 


0.647 


0.520 


16 


Lufthansa 


0.956 


1.190 


17 


British Air 


0.893 


0.974 


18 


Air France 


0.875 


1.089 


19 


Alitalia 


0.840 


1.250 


20 


SAS 


0.838 


1.289 


21 


KLM 


0.946 


1.098* 


22 


Swissair 


0.950 


1.360* 



The indirect cost per passenger (v) was computed as follows. The total indirect cost for 
an airline was computed by subtracting the total flight costs from its total cost. The total 
indirect cost for a flight segment was estimated by allocating the carrier's total indirect 
costs in proportion to the revenue generated from that particular route segment. Then, the 
per-passenger indirect cost for a flight segment (v) was computed by dividing the 
segment indirect cost by the segment passenger volume (Q). 

For the case of a Uvo-segment route, the unit cost is obtained by adding the two 
segments' unit costs per passenger. The carrier profit from a route is then obtained by 



13 



taking the difference between the fare and the route's unit cost, and multiplying it by the 
route demand volume. In this way, load factors are endogenized in the model. This has 
important implications for dynamic efficiency effects. Entry affects costs in two ways. 
First, entry of a low cost carrier affects incumbents ' costs by putting pressure on input 
prices and productivity. Second, it changes incumbents ' passenger volume, which, in 
tum, changes their per-passenger segment costs by being at a different point on the 
economies of density curve. 



4. Alternative Scenarios and Construction of the Base Case 

The model was applied to the cases of Canada-Japan (Vancouver-Tokyo market), 
Canada-Germany (Toronto-Frankfurt market) and Canada-Australia market. This paper 
reports only the empirical results for the Canada- Japan case. 

For each case, the following alternative scenarios are simulated and the equilibrium 
results are compared: 

(a) Base Case: price and capacity (frequency) regulated 

(b) Price regulation/capacity competition (with/without new entry) 

(c) Price competition/capacity regulation 

(d) Full competition (no restrictions on price, capacity or entry) 

At first, the simulation results were obtained for each of the above scenarios under the 
assumptions of differentiated monopolistic competition and closed bilateral trade (no 
alternative routing via third countries). Later, the selected simulation experiments were 
conducted to examine the effects of an alternative market structure: oligopoly vs. 
monopolistic competition and with and without explicitly accounting for traffic diversion 
to third country routes. 

The Base Case: Price, Entry and Capacity Regulated 

The key variables to describe the base case in an OD market are the airlines serving the 
market, type of aircraft being used, the passenger volume by airiine, fares by airline, 
frequency by airline, travel time by airline and a carrier specific preference (nationality) 
factor. Most of these data are collected from a variety of sources such as Transport 
Canada and US Department of Transportation, ICAO, Official Airiine Guide (OAG), 
travel agents and the airlines themselves. 

The carrier specific preference factor is computed by calibrating the demand model to the 
base case data. The difference between the observed passenger volume for a route-carrier 
combination and the predicted demand, i.e., proportion of the traffic volume that cannot 
be explained by the route-carrier characteristics, is regarded as the carrier specific 
preference factor. Table 2 illustrates the type of input variables needed for calibration of 
demand. 



14 



Table 2: Example Input Table for CALM Model 



BASE CASE ROUTE/CARRIER CHARACTERISTICS 


Carrier Plane Type PAX Fare Frequency Travel 
Code Code Code^ Time 


National 
preference 


1 1 63,875 434.30 10 7.50 
7 1 63,875 434.30 10 7.50 


1.00 
1.50 



Based on the review of empirical studies and surveys on demand elasticities,' the demand 
parameters assumed to range between the upper and lower bound values listed in Table 3. 

Table 3: Demand Parameters Used 



Probability Range* 

Lower Median Upper 
Demand Parameters 10% Value 10°/c 



'0 

Value Value 



Elasticity of Substitution 1.5 2.0 2.5 

Elasticity of Demand with Respect to Frequency 0.05 0.10 0.20 

Elasticity of Demand with Respect to Own Price -.075 -1 .35 -2.0 



"This probability range corresponds to an eighty percent confidence interval. 



5. Canada-Japan Results 

The Canada-Japan market was and is still being tightly regulated by the current bilateral 
agreement. This agreement controls carrier and route designation (including point of 
origin and point of access), seat capacity of aircraft used, and prices (single disapproval). 

In 1993, Canadian Airlines International (Canadian) and Japan Airlines (JAL) served 
Canada- Japan market. Canadian operated 8 flights per week between Vancouver (YVR) 
and Tokyo's Narita airport while JAL operated 7 passenger flights. Both carriers charged 
the median discount fare of CS680 per one-way passenger. Canadian and JAL carried 
99,720 and 88,050 passengers, respectively, in 1993. Based on our model Canadian 
made a profit of CS24 million on this route while JAL made only about CS2 million.'" 



See Oum, Waters and Yong (1992) for a sun-ey of transport demand elasticities, Oum, Gillen and Noble 
(1986) for estimates of elasticities of substitution, and Morrison and Winston for estimates of frequency 
elasticity' of demand. 

Although an additional routmg sen-ed by Canadian (Toronto - Narita) was included in our simulation, 
this paper reports the results on the Vancouver-Tokyo market onlv. 



15 



Effects of Removing Frequency Regulation: 

The market equilibrium results for the case of frequency competition (while regulating 
price and entry) are reported in Table 4, Column 2 (Without Entry). Under this scenario 
both carriers maintain profitability, and total route demand is expected to increase by only 
1 percent. However, relative to the base case this market outcome is less profitable for 
Canadian and slightly more profitable for JAL. The welfare gain to Japan is 
approximately $2 million dollars while the welfare loss to Canada is approximately $6 
million dollars, making a combined net welfare loss of $4 million. For Canadian, the 
reduction in profitability occurs because the bilateral change increases capacity at a rate 
that exceeds demand, pushing the segment load factor below 60 percent. This in turn has 
a negative impact on carrier costs and hence profitability. Canadian's position is further 
eroded by the observed carrier preference for Japanese carriers in the market. 

Table 4, Column 3 (With Entry) reports the equilibrium results for the case of removing 
both frequency and entry regulations while maintaining price regulation. In this scenario, 
a new carrier, namely, Air Canada enters the market with six flights per week. At 
equilibrium, each incumbent carrier offers six flights per week. The relative cost 
efficiency of these carriers places Canadian Airlines International (CAI) as the most cost 
competitive (and thus most profitable), followed closely by Air Canada and Japan 
Airlines. The entry of a new carrier in this market has a significant impact on route 
traffic volume - a 37 percent increase over the case of frequency competition without 
entry. This scenario also produces positive welfare impacts relative to the no entry case. 
The gain in aggregate profits for carriers is combined with the gain in consumer benefits 
of approximately S30 million to produce a total net welfare gain of $61 million dollars. 
Since the Japanese passengers dominate the market, non-Canadians capture the majority 
of the gain in consumer benefits while Canada captures a majority of the carrier profits 
due to the entry of Air Canada. 

Effects of Removing Price Regulation: 

Table 5 exhibits the equilibrium results of price competition with frequency regulation. 
The results with and without entry regulation are contained in the table. The results with 
entry regulation, Column 2, show that CAI reduces price by more than 27 percent while 
JAL reduces it by 19 percent. JAL can attract an almost equal number of passengers 
while charging a substantially higher fare than CAI because Japanese passengers prefer to 
fly with JAL (a positive carrier specific factor). Consumers now, in part, captvu-e the 
economic rents previously captured by the carriers. The net welfare effect is positive, 
with the consumer benefits slightly outweighing the reduction in carrier profits. 

Column 3 (With Entry) of Table 5 reports the (equilibrium) results for the case of 
removing both price and entry regulations while keeping frequency regulation. The 
results show that the entry of a new carrier (Air Canada) is important not only for 
stimulating market demand, but also because it offers consumers a wider range of choice. 
Consumer benefits are far in e.xcess of the scenario where frequency competition takes 



16 



place in the absence of price competition. It is also true that consumer benefits increase 
largely via a transfer of carrier profits; net welfare has more than doubled as compared to 
the case of price competition without removing frequency and entry regulations. Unlike 
the previous scenario, the distribution of benefits falls more heavily in favor of Canada 
because a Canadian carrier enters a profitable market. This is in stark contrast to the no 
entry case, where the Canadian carrier suffers and Canadian consumer benefits are much 
smaller. 

Effects of Removing Regulations on Price, Frequency and Entry: 

This scenario is essentially the bilateral open-skies agreement that does not involve 
opening of the S"" fi-eedom traffic rights. The simulation results are reported in Table 6. 
In this scenario, Air Canada enters the Vancouver- Tokyo market, and each of the three 
carriers would serve flights per week. The entry of a new carrier (Air Canada) makes the 
market significantly more competitive, and as a result, the equilibrium prices are 
significantly lower than the case with entry regulation. As a result of both entry and lower 
prices traffic volume would increase by about 50%. Furthermore, there would be a five- 
fold increase in the welfare gain to Canada because of the entry of Air Canada while 
Japan's welfare gain is limited to an increase fi-om S23 million to $38 million. 

Canada-Japan Outcome under an Oligopoly Market Structure 

Oligopoly firms can exercise market power by erecting entry barriers and charging 
substantial market-up over marginal cost. Oum, Zhang and Zhang (1993) have found that 
most airlines play a Coumot game in their markets. At the Coumot-Nash equilibrium an 
airline with a large market share can charge a substantial mark-up over and above their 
marginal costs. 

Table 7 compares the equilibrium outcomes of the base case and the two cases of the 
price and firequency competition case (the case of monopolistic competition and the case 
of Coumot oligopoly). In these simulations we assume that the incumbents are 
successftil in blocking entry of potential competitors. The aggregate gain to consumers is 
approximately S32.7 million for the case of price and fi-equency competition without any 
oligopoly markup. With the imposition of the mark-up the consumer gain is reduced to 
$13.0 million because of the increased prices and reduced volumes. The carriers' total 
profits increase over the no-mark-up case by $9.33 million. The aggregate net welfare 
shrinks fi-om $38 miUion for the no-mark-up case to $28.3 million for the oligopoly 
markup case. 

Incorporation of the Aspects of Traffic Diversion to Third Country Routes: " 

As we discussed previously, a bilateral liberalization not only increases competition in 
the direct routes, but also induce those who are traveling via foreign cities to return to the 



" In a closely related work, Dresner and Oum (199S) investigates the effects of Canada's "facilitating" and 
US liberal bilateral air agreements on the share of \isitors rravelling directly to Canada, as opposed to 
transitins through the United States. 



17 



direct routes. The liberalization can improve the relative attractiveness of the direct flight 
as compared to travel via a third country point. For example, the liberalization of 
Canada- Japan bilateral agreement would induce the passengers traveling between Canada 
and Japan via a U.S. point to use the direct route. Furthermore, it would cause some of 
the U.S. and Japanese passengers traveling between the U.S. and Japan to route their 
travel via Canadian points. Intuitively, the effects of accounting for such traffic diversion 
in the model is expected to increase both carriers' and consumers' benefits. These added 
passengers benefit the airlines in two ways. First, they increase traffic density on the 
Vancouver - Tokyo route, and thus, reduce per-passenger cost. Second, the added 
demand increases the market clearing price slightly, further increasing revenue and profit 
margins. 

Table 8 reports the simulation results for the case involving third country routing. For 
the Vancouver - Tokyo market, traffic diversion occurs when passengers who now fly 
between Vancouver and Narita through US gateways may consider a direct Vancouver 
routing. Our results show that the removal of price and frequency regulation in the 
Canada-Japan bilateral is likely to reduce the passengers who travel via Vancouver- 
Seattle-Tokyo (via Northwest) from 11,300 to 8,950 persons. Likewise, it will reduce 
those who travel via Vancouver-San Francisco-Tokyo (via United) from 7,520 to 5,520 
persons. These translate into additional consumer benefits of approximately $1 million 
dollars or 3 percent of the original benefits estimate. Similarly, both Canadian's and 
JAL's profit increases. 



Summary Results on the Canada-Japan Case 

The simulation results on the Canada- Japan case can be summarized as follows: 

• When price is regulated, frequency competition benefits both countries only if entry 
regulation is also removed. Frequency competition without freeing the entry neither 
increase airline profits nor improve consumer welfare. 

• For the Canada- Japan case, frequency competition with entry freedom (when price is 
regulated) increases Canada's welfare more than Japan's because the new entrant is a 
Canadian carrier, Air Canada. 



• 



When price regulation is removed while keeping frequency regulation intact, both 
carrier profits and consumer benefits increase substantially when frequency is 
regulated at reasonable level. The total consumer surplus increases more to Japanese 
passengers than to Canadian passengers because a large majority of the passengers on 
Vancouver-Tokyo segment are Japanese nationals. Needless to say, both consumer 
surplus and carrier profits would be significantly affected if frequency is regulated at 
wrone value. 



's 



The benefits of price competition get more than doubled if entry is also freed. 
Although Air Canada is the only carrier expected to enter the market, .the overall 



welfare gain is greater for Japan than for Canada because of the dominance of 
Japanese passengers on Vancouver-Tokyo market. 

The complete liberalization of pricing, frequency and entry leads to the welfare 
maximizing market outcome. 

Oligopoly solution (Coumot Nash equilibrium) increase carrier profits substantially 
while reducing consumer surplus. 

The effects of liberalization of price and frequency regulations on 0-D traffic volume, 
carrier profits and consumer surpluses are greater when the model takes into account 
of the third country routing possibilities. 



6. Summary and Conclusion 

In this study, we attempted to develop a model with which to measure the economic 
effects of liberalizing bilateral air agreements between two countries. Our model allows 
us to measure not only the changes in equilibrium outcomes and welfare consequences of 
liberalizing a bilateral air transport agreement, but also the distribution of the gains and 
losses to carriers and consumers of each bilateral country and those of the third foreign 
countries. In particular, our model allows to measure the effects of changes in a bilateral 
agreement on the amount of traffic diversion between the direct bilateral routes and the 
indirect routes via a third country. We also provide an extension of our model to a case 
of oligopoly market outcome (Coumot Nash equilibrium). 

Since quality of services is important for determining air transport demands, costs and 
consumer welfare, our main model is developed by adapting the monopolistic 
competition model of Spence (1976) and Dixit and Stiglitz (1977) to the air transport 
situation. This allowed us to incorporate the attributes of service quality such as 
frequency of service, travel time, number of connections required to complete a travel, 
and national flag carrier preference factor in the demand model. We adopted the 
"Armington" assumption by specifying our Origin-Destination specific demand model in 
the CES form and thereby treating the 'route-carrier' combinations serving an Origin- 
Destination market as imperfect substitutes to each others. Quality aspects are treated in 
the framework of hedonic price theory by specifying the quality-adjusted price (quantity) 
as a multiplication of the observed price (quantity) by the reciprocal quality index 
function (the quality index function). 

The total cost of a flight segment consists of the costs that vary with flight frequency and 
those that vary v/ith number of passengers carried. This implies that our model allows 
the carriers to adjust their unit costs dynamically with the traffic density on the route 
segment. 



Since it was not possible to obtain closed form expression for equilibrium solutions, 
numerical simulations were conducted to measure the effects of changing the following 
major policy levers in a bilateral air transport agreement: 

• Removing price regulation while retaining frequency and entry restrictions 

• Removing price and entry regulation while retaining frequency restrictions 

• Removing frequency regulations while retaining price and entry regulations 

• Removing frequency and entry regulations while retaining price regulation 

• Removing price and frequency regulations while retaining entry restriction 

• Removing all price, frequency and entry regulations (de facto, open skies) 

Our model was applied to the cases of Canada- Japan, Canada-Germany, and Canada- 
Australia bilateral agreements. Although this paper reports the empirical results on the 
Canada-Japan bilateral case only, they are by and large consistent with those of the 
Canada-Germany and Canada-Australia cases. 

Our key results can be summarized as follows: 

• Frequency competition without freeing entry or price regulation neither increase 
airline profits nor improve consumer welfare. Frequency competition with entry 
freedom increases the welfare of the nation whose carrier enters the market, i.e., the 
nation with lower cost carriers 

• Pricing freedom with frequency regulation increases the welfare of the nation with a 
larger share of passengers on the bilateral markets more than other countries. The 
benefits of pricing freedom are significantly affected by the regulated frequency of 
services. The benefits of price competition becomes more than doubled if entry is 
also freed. 

• Overall, allowing entry of new carriers increase the overall welfare the most, followed 
by the price freedom. Just the removal of frequency restrictions has the least effect on 
consumer welfare. 

• The complete liberalization of pricing, frequency and entry leads to the welfare 
maximizing market outcome. 

• Oligopoly solution (Coumot Nash equilibrium) increase carrier profits while reducing 
consumer surplus substantially. 

• The effects of liberalization of price and frequency regulations on 0-D traffic volume, 
carrier profits and consumer surpluses are greater when the model takes into account 
of the third country routing possibilities. 



20 



Our current research attempted to measure the effects of Hberahzing the bilateral 
agreement with a single country. We have not attempted to measure the effects when a 
country hberalizes its bilateral agreements with many countries as the U.S. government is 
pursuing. Extending our model to handle such a situation would not straight forward, 
but it is an interesting avenue for future research. 

Our analysis is also limited to measuring the effects on the producers and consumers of 
air transport services only, ignoring other benefits of bilateral air liberalization including 
the benefits to tourism sector. Certainly, there is a need to incorporate several related 
sectors including tourism in the analysis of air transport matters. However, use of a full 
general equilibrium model for air transport analysis may not be an effective avenue to 
pursue. Since air transport sector, especially each bilateral air transport market, is small 
relative to other sectors of the economy, it would be difficult to identify the effects of 
liberalization of a small number of bilateral agreements within a full general equilibrium 
model because those small effects are likely be buried in the changes in larger economic 
sectors. 



21 



References 

^Zl''^\^rZr:^ ^^°^ -''''--' '- ^-^-^ ^^^^-^"^^^^^ 'y ^1- of 

Caves, aw., L. R. Christensen, and W.E. Diewert (1982). "Multilateral Comparisons of Output 
Input, and Productivity Using Superlative Index Numbers." Economic Journal, 92, 1982, pp. 73- 

^^.tLu^iSfp^-.;^^^ ^""^ ' ''''''■■ ^' ^''^^^ •" ^'-^ ^--^- 

Dempsey, Paul Stephen (1987). "Turbulence in the 'Open Skies': The Deregulation of 
International Air Transport", Transportation Law Journal Vol. 15, 1987, pp. 325-42. 

Dixit. A.K. (1992). "Optimal Trade and Industrial Policies for the U.S. Automobile Industry " in 
Press) °^^™^" ^w/'e^ec/ Competition and International Trade, (Cambridge, Mass.: MIT 

Dixit A. and J. Stiglitz (1977), "Monopolistic Competition and Optimal Product Diversity" 
Amencan Economic Review 67: 297-308. ^'vciMiy , 

Doganis, Rigas (1993). "The Bilateral Regime for Air Transport: Current Position and Future 
Prospects in International Air Transport: The Challenges Ah^.H Paris: OECD Publications, 

Dresner, M. and T. H. Oum (1998), "The Effect of Liberalized Air Transport Bilaterals on 
Vol 3^, Part s's^pTs ^-33?"' ^''' °^ ^'"''^''"' J»""^^^ "^ ^^'^^^^-^ Econon.,.. ..H P.i;.y 

Gillen, D.W., T.H Oum and M. Tretheway (1990). "Airiine Cost Structure and Policy 
Implications: A Multi-Product Approach for Canadian Airlines," Journal of Transport 
Economics and Policy, vol. 24: 9-34 ^'un^pun 

Harris R. (1984), "Applied General Equilibrium Analysis of Small Open Economies with Scale 
Economies and Imperfect Competition", American Economic Review, vol. 74. no.5 J 016-32. 

No'^;wts,;8?pj'7M!^r'°°''" """"'"• ^""""" '"-" °-^^'°"°™"°' ™' 

Harris, R. (1991). "Market Access in hitemational Trade: A Theoretical Appraisal" in R Stem (ed ) 
U.S. Canadian Trade and Investment Relations yvitli Japan (University of Chicago Press), 1991 

Harries, R. and D^Cox. (1992). "North American Free Trade and Its Implications for Canada- 
Results from a CGE Model of North American Trade," 77ie World Economy, vol. 15: 31-44. 

Harris R. (1993). "Trade and Industnal Policy for a 'Declining' Industry: the Case of the U S 

,lml? .^/ ' '" ^- ^^g™^" ^"^ A. Smith (eds.) Empirical Studies of Strategic Trade Policy, 
(NBER and University of Chicago Press). 



T) 



Helpman, E. and Paul Krugman (1985). Market Structure and Foreign Trade: Increasing 
Returns, Imperfect Competition, and the International Economy (1985) MIT Press, Cambridge 
MA 

Hufbauer, Gary and Christopher Findlay, eds, (1996). Flying High: Liberalizing Civil Aviation in 
the Asia Pacific, Washington, Institute for International Economics, Washington, 1 996 

Hurdle, G.J. et al.(1989). "Concentration, Potential Entry, and Performance in the Airline 
IndMstr^" Journal of Industrial Economics, vol. 38: 119-139. 

Morrison, S., and C. Winston (1986), Economic Effect of Airline Deregulation, The Brookings 
Institution, Washington,D. C. 

Oum, T.H., D.W. Gillen and S.E. Noble (1986), "Demands for Fareclasses and Pricing in Airline 
Markets", Logistics and Transportation Review, vol. 22, no.3: 195-222 

Oum, Tae H. and Yu, C. (1995). "A Productivity Comparison of the World's Major Airlines," 
Journal of Air Transport Management, vol.2, No.3/4 (1995), pp. 181-195 

Oum, Tae H. and C. Yu. (1998). "Cost Competitiveness of the World's Major Airlines: An 
International Comparison, Transportation Research A. vol.32, no.6, 407-422. 

Oum, Tae H., W.G. Waters and Jong-Say Yong, (1992). "Concepts of price Elasticities of 
Transport Demand and Recent Empirical Estimates", Journal of Transport Economics & Policy 
(May) 1992 pp: 139-154 

Spence, Michael (1976). "Product Selection, Fixed Cost and Monopolistic Competition", Review 
of Economic Studies 43:21 7-236 

Street, John, David Smith and Scott Savage (1994). "An Analysis of the Trade-Offs in 
International Aviation", 19th Australian Transport Research Forum, 1994 

Tirole, Jean (1988), The Theory of Industrial Organization, MIT Press. 

Transport Canada (1994). "Canada's International Air Policy", TP12276, December 1994. 

Windle, Robert (1991). "The World's Airlines: A Cost and Productivity Comparison." Journal 
of Transport Economics and Policy, 1991. 

Windle and Dresner (1992). "Partial Productivity Measures and Total Factor Productivity in the 
Air Transport Industry: Limitations and Uses." Transportation Research, Vol. 26, 1992. 



Table 4: Vancouver - Tokyo: Frequency Competition /Price Regulation 



Median Discount Fare (One Way in $) 
Canadian 
Japan Aii Lines 
Air Canada 

Average Weekly Frequency 
Canadian 
Japan Air Lines 
Air Canada 

Demand (Thousands of One Way Passengers) 
Canadian 
Japan Air Lines 
Air Canada 
Total Demand 

Profits (Millions of $) 

Canadian 

Japan Air Lines 

Air Canada 
Total Profit 

Welfare Impacts* 

Consumer Benefits - aggregate (Millions of $) 
Consumer Benefits to Canada (Millions of $) 
Consumer Benefits to others (Millions of $ 

Producer Benefits (Millions of $) 
Chg. in Canadian Profit (Millions of $) 
Chg. in Japan Airlines Profit (Millions of S) 
Chg. in Air Canada Profit (Millions of $) 
Chg. in Total Profit 

Aggregate Welfare Gain (Millions of S) 
Aggregate Welfare Gain to Canada (Millions of $) 
Aggregate Welfare Gain to others (Millions of S) 



Base Case* No Entry With Entry 



*A11 results reported in 1993 Canadian dollars. 



680 


680 


680 


680 


680 


680 
680 


8 


8 


6 


7 


8 


6 
6 


99.7 


87.6 


80.5 


88.1 


102.2 


89.2 
89.2 


187.8 


189.8 


258.9 


23.9 


17.9 


21.8 


2.0 


3.2 


10.4 
30.2 


25.9 


21.1 


62.4 




1.1 


29.9 




0.3 


8.9 




0.8 


20.7 




-6.1 


-4.5 




1.2 


5.4 
30.1 




-4.9 


31.0 




-3.7 


60.6 




-5.7 


34.5 




2.0 


26.1 



24 



Table 5: Price Competition /Frequency Regulation 



Vancouver - Tokyo 
(Frequency regulated at 7 flights per week) 



Median Discount Fare (One Way in $) 
Canadian 
Japan Air Lines 
Air Canada 

Demand (Thousands of One Way Passengers) 
Canadian 
Japan Air Lines 
Air Canada 
Total Demand 

Profits (Millions of $) 

Canadian 

Japan Air Lines 

Air Canada 
Total Profit 

Welfare Impacts * 

Consumer Benefits - aggregate (Millions of $) 
Consumer Benefits to Canada (Millions of $) , 
Consumer Benefits to others (Millions of $ 

Producer Benefits (Millions of S) 
Chg. in Canadian Profit (Millions of S) 
Chg. in Japan Airlines Profit (Millions of $) 
Chg. in Air Canada Profit (Millions of S) 
Chg. in Total Profit 

Aggregate Welfare Gain (Millions of $) 
Aggregate Welfare Gain to Canada (Millions of $) 
Aggregate Welfare Gain to others (Millions of S) 
•All results reported in 1993 Canadian dollars. 



Base Case* No Entry 



With Entry 



680 


546 


505 


680 


601 


560 
513 


99.7 


115.6 


115.5 


88.1 


115.7 


115.4 
115.5 


187.8 


231.3 


346.5 


23.9 


25.6 


19.5 


2.0 


13.3 


5.8 
19.0 


25.9 


38.9 


44.3 




21.8 


50.3 




6.6 


15.4 




15.3 


34.5 




-0.7 


-4.4 




8.4 


3.8 
19.0 




7.8 


18.0 




29.7 


68.2 




5.9 


30.0 




23.8 


38.3 



25 



Table 6: Price and Frequency Competition 
(Vancouver - Tokyo) 



Final Equilibrium 



Median Discount Fare (One Way in $) 
Canadian 
Japan Air Lines 
Air Canada 

Frequency 
Canadian 
Japan Airlines 
Air Canada 

Demand (Thousands of One Way Passengers) 
Canadian 
Japan Air Lines 
Air Canada 
Total Demand 

Profits (Millions of $) 
Canadian 
Japan Air Lines 
Air Canada 
Total Profit 



Welfare Impacts 

Consumer Benefits - aggregate (Millions of $) 
Consumer Benefits to Canada (Millions of $) 
Consumer Benefits to others (Millions of S) 

Producer Benefits (Millions of $) 
Chg. in Canadian Profit (Millions of $) 
Chg. in Japan Airlines Profit (Millions of S) 
Chg. in Air Canada Profit (Millions of $) 

Chg. in Total Profit 

Aggregate Welfare Gain (Millions of S) 
Aggregate Welfare Gain to Canada (Millions of $) 
Aggregate Welfare Gain to others (Millions of S) 
*A11 results reported in 1993 Canadian dollars. 



Base Case Without 
Entry 



680 
680 



99.7 
88.1 

187.8 



23.9 
2.0 

•25.9 



494 
549 



131.9 
131.9 

263.8 



23.3 
8.5 

31.8 



32.7 

9.8 

22.9 



-0.6 
6.6 

5.9 

38.6 

9.2 

28.8 



With Entry 
Freedom 



505 
560 
513 



7 
7 
7 



115.5 
115.4 
115.5 
346.5 



19.5 

5.8 

19.6 

44.2 



50.3 
15.4 
34.5 



-4.4 
3.8 
19.0 
18.0 

68.2 
30.0 

38.3 



26 



Table 7: Price/Frequency Competition with and without Oligopoly Markup 
(Vancouver - Tokyo) 



Base Case without With Oligopoly 

oligopoly mark- Mark-Up 



Median Discount Fare (One Way in $) 








Canadian 


680 


494 


523 


Japan Air Lines 


680 


549 


575 


Average Weekly Frequency 








Canadian 


8 


8 


7 


Japan Air Lines 


7 


8 


7 


Demand (Thousands of One Way Passengers) 








Canadian 


100.0 


131.9 


120.2 


Japan Air Lines 


88.1 


131.9 


123.3 


Total Demand 


187.8 


263.8 


243.4 


Profits (MilUons of $) 








Canadian 


23.9 


23.3 


26.3 


Japan Air Lines 


2.0 


8.5 


14.9 


Total Profit 


25.9 


31.8 


41.1 




Welfare Impacts 


Consumer Benefits - Aggregate (SM) 




32.7 


13.0 


Consumer Benefits - Canada ($M) 




9.8 


3.9 


Consumer Benefits - Others ($M) 




22.9 


9.1 


Producer Benefits 








Chg. in Canadian Profits 




-0.6 


2.4 


Chg. in JAL Profits 




6.6 


12.9 


Change in Total Profits (SM) 




5.9 


15.2, 


Aggregate Welfare Gain (SM) 




38.6 


28.3 


Aggregate Welfare Gain to Canada (SM) 




9.2 


6.3 


Aggregate Welfare Gain to Others ($M) 




28.8 


22.0 



27 



Table 8: Price and Frequency Competition with Third Country Routings 



(Vancouver - Tokyo) 

Base Case 



Median Discount Fare (One Way in $) 
Canadian 
Japan Air Lines 
Northwest via Seattle 
United via San Francisco 

Average Weekly Frequency 
Canadian 
Japan Air Lines 
Northwest via Seattle 
United via San Francisco 

Demand (Thousands of One Way Passengers) 
Canadian 
Japan Air Lines 
Northwest via Seattle 
United via San Francisco 
Total Demand 

Profits (Millions of $) 

Canadian 

Japan Air Lines 
Total Profit 



Welfare Impacts 



Consumer Benefits - Aggregate ($M) 
Consumer Benefits - Canada ($M) 
Consumer Benefits - Others ($M) 

Producer Benefits 
Chg. in Canadian Profits 
Chg. in JAL Profits 
Change in Total Profits (SM) 

Aggregate Welfare Gain (SM) 
Aggregate Welfare Gain to Canada (SM) 
Aggregate Welfare Gain to Others (SM) 
* no change 



680 
680 
550 
550 



8 

7 
7 
7 



100.0 

88.1 

11.3 

7.5 

206.6 



23.9 

2.0 

25.9 



without 
Diversion 



494 
549 

NC* 
NC* 



8 
8 

NC* 
NC* 



131.9 
131.9 

NC* 
NC* 
263.8 



23.3 

8.5 

31.8 



32.7 

9.8 

22.9 



-0.6 

6.6 

5.9 

38.6 

9.2 

28.8 



with 
Diversion 

500 
554 
NC* 
NC* 



8 
8 

NC* 
NC* 



133.8 

134.5 

9.0 

5.5 

282.8 



24.9 
11.2 
36.1 



33.5 
10.1 
23.5 



1.0 
9.2 
10.3 

43.8 
11.1 
32.7 



28 



HUBBING AND HUB-BYPASSING 
Network Developments JDie Deregulated Eurojpean Airport Systan 



Jaap de Wh/FeterUitteiibogsart/rbaltcta Wei-Yun 

Civil Aviation De(>ar(ineat of ^e Neth^dauds 

•^Aviation Economics Section- 



Faper to b? presented at the ATRG conference, 
Hong Kong, June l^S^ 



Hubbing and hub-bypassing 

Introduction 

Most European national airlines were operating radial networks already decades before the 
U.S. domestic airline market was deregulated and hubbing became the major trend in this 
market. However, this radial network structure in Europe resulted from international market 
regulation, i.e. third and fourth freedom routes between the national home base and points 
abroad. Since at that time most European airlines hardly provided any connectivity between 
these third and fourth fi-eedom routes, most national airports in Europe hardly functioned as 
hubs^ during the pre liberalisation stage. 

An exception can be made for a few carriers, like for example KLM, Swissair and SAS, which 
only had small domestic markets. These carriers needed additional European feed from outside 
their own domestic market for their long-haul wide-body operations. Therefore in addition to 
their domestic O&D market airlines generated a transfer market through a scheduled 
connectivity between long-haul intercontinental and short-haul European/domestic flights. 
However the relatively limited number of ICA destinations hardly generated any network 
multiplier effect. The so-called gateway system usually focused on only two banks: for ICA 
arrivals and ICA departures, respectively. 

During the stepwise liberalisation of the EU air transport market various European airlines also 
started a hubbing system at their national home base next to this gateway function. At several 
national airports the connectivity between European arrivals and departures substantially 
improved through an increasing number of daily connection banks which can serve more 
destinations at higher frequencies. Short-haul to short-haul hubbing apparently took off inside 
Europe after the liberalisation of route and market entry. Compared to the U.S. domestic 
market the scale of hubbing has remained limited until now. The limited average travel distance 
inside Europe and the higher density of public transport and road networks mainly explain this 
difference. 

It can be expected that an increasing number of European hubs will get involved in the next 
stage of network developments, i.e. multiple Euro hubbing through alliances between 
European carriers, like Lufthansa and SAS, Swissair and Sabena, BA and Iberia, KLM and 
Alitalia. 

As a result of these hubbing developments and the extra transfer demand generated by it, 
congestion has exacerbated at most major airports in Europe. This growing congestion and the 
related noise problems at hub airports evokes a new discussion about the utility of providing 
substantial airport and noise capacity to foreign transfer passengers, thereby depositing the 
external noise effects of their travelling on the neighbouring area around the airport. Especially 
in the Netheriands hubbing is increasingly questioned nowadays by public interest groups in the 
context of long-term airport planning. Environmentalists increasingly characterise hub&spoke 
systems as inefficient systems by referring to the superfluous diversion within these networks 
compared to point to point connections. This seems to be a sufficient argument to announce 
the end of the hubbing era. The new era would bring a plethora of hub-bypassing routes 
between smaller uncongested airports that replace the indirect routes through the hub. 
Although the consequences of a decentralised point to point system in Europe seems to be 
rather unfavourable to the environment, this prophecy seems also to be rather unlikely from a 



' .\ hub is defined as an airport where the dominant and usually home-based carrier schedules it departing and 
arriving flights in short consecutive periods (banks or vva\es). and a transfer system provides an acceptable 
connectivity between arriving and departing flights. 



network-operational point of view. Theoretical arguments against this prophecy are clear and 
can be based on generalised costs of diversion time, extra transfer time, reduced waiting time 
and lower fares, compared to the alternative of a direct route, if available anyway. (See also 
Tretheway and Oum, 1992). However the discussion is also a little complicated by the fact that 
hubbing and hub-bypassing are two simultaneous network characteristics, which can be 
explained by the same arguments. Geographical characteristics (route length) as well as the 
actual route densities and cost competitiveness ^ determine whether new entrants are able to 
provide a direct connection between two spoke points. All in all however, the general 
arguments point in the direction that the future picture of network developments in Europe 
will be increasingly dominated by hub and spoke systems. 

As a contribution to the debate on hubbing versus hub-bypassing we have analysed actual 
network developments between categories of airports in Europe during the last two decades. 
In this analysis we used the unique collection of ABC/0 AG-data for the years 1984, 1990, 
1993 and 1997, available in the database of the Dutch Civil Aviation Department. This paper 
contains the findings of our analysis. 

Hypotheses regarding netyvork developments within the European airport system 

First of all we describe a classification of airport categories and route types. This classification 
provides the basic data for determining whether the hub&spoke phenomenon or the hub- 
bypassing phenomenon was the dominating trend in the period from 1984 until 1997. More 
explicitly, the following hypotheses with respect to the hub-bypassing and the hub&spoke 
phenomenon respectively, were examined. 

If the hub-bypassing phenomenon had been the dominating trend, then: 

• the percentage of connections between European regional airports would have increased; 
and/or 

• the percentage of frequencies offered between European regional airports would have 
increased; and/or 

• the percentage of seats offered between European regional airports would have increased. 

In reverse, if the hub&spoke phenomenon had been the dominating trend, then: 

• the percentage of connections between European regional airports and the hub airports 
would have increased; and/or; 

• the percentage of frequencies offered between European regional airports and the hub 
airports should have increased; and/or 

• the percentage of seats offered between European regional airports and the hub airports 
would have increased. 

Classification of European Airports 

All in all, five different airport size categories have been defined a little arbitrarily. Each 
category allows a minimum and maximum numbers of available seats offered on both intra- and 
intercontinental routes to and from the airport involved. The classes were calibrated on the 
data for the reference year 1990. For the other years, viz. 1984, 1993 and 1997, the various 
classes were scaled according to the average market growth. As a result, Tablel provides an 
overview of the comparable airport size classes for each of the four years. Data used in the 
classification were derived from the OAG/ABC timetables for a representative week in July 



" Price discriiniuation through revenue uiaimgenient is an important tool for the incumbent carrier to counter 
this new competition. 



To transpose these weekly figures into comparable yearly figures, the rule of thumb was used 
that an annual total is roughly 48 times the representative week in July. 



Table 1: Oassincation of European airports according to seat capacity offered on scliediiled passenger lliglits 



Category 




cap/wectc 


cap/year 


cap/weel< 


cap/year 


capAvceIc 


cap/year 


cap/week 


cap/year 






1984 


1984 


1990 


1990 


1993 


1993 


1997 


1997 


Very Large 


>- 


175.000 


8,4 mill 


250.000 


12,0 mill 


281.250 


13.5 mill 


350.000 


16,8 mill 


Large 


>- 


70.000 


3,4 mill 


100.000 


4,8 mill 


112.500 


5,4 mill 


140.000 


6,7 mill 


Medium 


>- 


7.000 


340 thd 


10.000 


4gOt]id 


11.250 


540 thd 


14.000 


670 thd 


SmaU 


>- 


1.750 


g4thd 


2.500 


120 thd 


2.813 


135 thd 


3.500 


170 thd 


VerySmaU 


< 


1.750 


84 thd 


2.500 


120 thd 


2.813 


135 thd 


3.500 


170 thd 



Table 2 shows the number of European airports assigned to each of the five categories in each 
of the four years. The conclusion seems to be justified that the number of European airports in 
the various categories has remained rather stable during the period analysed. However, the 
exceptional case is the number of European airports in the category ^ery small". Between 
1984 and 1990, this number increased by more than a 100, a relative growth of about one 
third. This significant increase is probably correlated with the gradual liberalisation of 
European aviation. Cross-border interregional aviation within the EU had already been 
liberalised in 1983. From this moment on, regional airlines were free to start interregional 
services between secondary and/or tertiary airports with a maximum aircraft size of 70 seats. 
As a consequence an increasing number of smaller general aviation airports were served by 
cross-border scheduled passenger services. From 1987 on, all aviation in the EU has gradually 
been liberalised in three consecutive steps. In 1997 the liberalisation of the EU air transport 
market was completed when cabotage was fiilly allowed. 

Table 2: Number of European airports according to seat- capacity class for scheduled passenger services 



Category 


1984 


% 


1990 


•/. 


1993 


% 


1997 


% 


Very Large 


>- 


5 


0.9 % 


4 


0,6 % 


5 


0.7 % 


5 


0.8 % 


Large 


>- 


12 


2.2 % 


15 


2,3 % 


13 


1.9 % 


13 


2.3 % 


Medium 


>- 


97 


17.5 % 


95 


14,4% 


96 


14,3 % 


99 


15,0% 


SmaU 


>— 


151 


27.3 % 


140 


21,2% 


152 


22.7 % 


133 


20.1 % 


Very SmaU 


< 


289 


52.2% 


405 


61.5% 


405 


60,4 % 


408 


61.8% 


Total 




554 


100% 


659 


100 % 


671 


100 % 


660 


100 % 



Source: GAG/ABC 

Based on the aforementioned classification of European airports, 

• the number of intra-European scheduled passenger routes, 

• the number of frequencies offered on these intra-European routes, as well as 

• the seat capacity offered on these intra-European routes, 

were selected from the OAG/ABC time schedules for the respective years in the analysis. 



Classification of route types 

To analyse network developments between the different airport categories, three different 
route types can be distinguished among European airports: 



• interregional connections, i.e. connections offered between or within the airport categories 
very small, small and medium; 

• interhub connections, i.e. connections offered between or within the airport categories large 
and very large; 

• hub&spoke connections, i.e. connections offered between the airport categories very small, 
small and medium on the one hand and the airport categories large and very large on the 
other. 



Historical developments in numbers of connections 

Table 3 shows that the number of interregional routes was slightly below the average market 
growth rate for the period of 1984-1990 and slightly above the average market growth rate for 
the period of 1990-1997. All in all, there has not been a significant shift in the distribution of 
intra-European connections over the route types interregional, interhub and hub&spoke. On 
average one may conclude that the distribution according to route type has been fairly stable. 
The development of the number of intra-European connections does not render any convincing 
evidence for either the hypothesis that there has been an intensification of interregional traffic 
in Europe or the hypothesis that there has been a strong development towards a hub&spoke 
network structure in Europe. 



Tabic 3: Number of tntra-European scheduled passenger connections according to route type 



Cateeory 


198-4 


% 


1990 


% 


1993 


% 


1997 


% 


Interhub connections 


116 


3,7% 


154 


3,6% 


123 


4,1% 


172 


4,8% 


Average annual gnnvth •/. 






4.8V. 


1984-1990 


-7.2% 


1990-1993 


8.7% 
1.6% 


1993-1997 
1990-1997 


Hub&Spoke connections 


790 


39,0% 


1078 


39.4% 


1170 


39,3% 


1403 


38,8% 


Average annual gmwth */• 






5.3% 


1984-1990 


2.8 % 


1990-1993 


4.6% 
3.8% 


1993-1997 
1990-1997 


Interregional connections 


1117 


35,2% 


1303 


33,0% 


1687 


56.6% 


2039 


56,4% 


Average annual growth */• 






5.1% 


1984-1990 


3.9% 


1990-1993 


4.9% 
4,5% 


1993-1997 
1990-1997 


Total 


2023 


100,0% 


2733 


100.0% 


2980 


100.0% 


3614 


100,0°i 


Average annual growth */• 






5.2% 


1984-1990 


2.9% 


1990-1993 


4.9% 

4.1% 


1993-1997 
1990-1997 



Source: GAG/ABC 

Historical developments in frequency levels and seat capacity 

Tables 4 and 5, on the contrary, show a major shift in the distribution of both frequencies and 
seat capacity offered on interregional, interhub and hub&spoke connections. The share of 
interregional traffic in overall totals has dropped significantly, both in frequencies and in seat 
capacity. This highlights the fact that interregional traffic is losing ground especially to 
hub&spoke traffic. The share of interhub traffic is fairly stable, in terms of frequencies as well 
as in terms of seat capacity. 



We therefore conclude that the findings for the period of 1984 - 1997 show an increased hub- 
orientation of the regional airports, instead of an increased orientation towards each other. In 
other words, hubbing has substantially increased in the European market, whereas the contrars' 
is true for hub-bypassing in the EU during the period analysed. 



Tabic 4: Number ofwcckly intra-Europcui scheduled Trequenclcs" for the various route types 



Category 


198.1 


% 


1990 


% 


1993 


•/. 


1997 


% 


Interhub connections 


3120 


13,4% 


4832 


14,1% 


5044 


13,1% 


7021 


14.5% 


Aveni{e annual growth */• 






7,6% 


1984-1990 


1.4% 


1990-1993 


8.6% 
5.5% 


1993-1997 
1990-1997 


Ilub&Spoke conncctloiu 


9923 


42,5% 


15445 


45,1% 


18042 


46,9% 


23833 


49,2% 


Average annua] growth ■/• 






7.7% 


1984-1990 


5J% 


1990-1993 


7.2% 
6.4% 


1993-1997 
1990-1997 


Interregional connections 


10283 


44,1% 


13938 


40,7% 


15362 


40,0% 


17630 


36,4% 


Average annual growth ■/• 






5.2V. 


1984-1990 


3.3% 


1990-1993 


3.5% 
3.4% 


1993-1997 
1990-1997 


Total 


23328 


100,0% 


34215 


100,0% 


38448 


100,0% 


48484 


100,0% 


Average annual growth Va 






fi,6y.% 


1984-1990 


4,0% 


1990-1993 


6.01'. 
5.1% 


1993-1997 
1990-1997 



1) Frequency: a return flight, l.e. an outgoing and on incoming aircraft movement 
Source: OAG/ABC 



Table S: Total seat capacity" olTered on a weekly basis Tor the various route types in intra-European scheduled passenger traRIc 



Cafeeory 


1984 


•/. 


1990 


% 


1993 


% 


1997 


•/. 


Ijiterfaub connections 


914 thd 


20,5% 


1430 thd 


20.7% 


1446 thd 


18,5% 


1990 thd 


20,3% 


Average annua] growth */• 






7.7% 


1984.1990 


0.4% 


1990-1993 


8.3% 
4.8% 


1993-1997 
1990-1997 


Hub&Spoke connections 


2073 Ihd 


46,4% 


3365 thd 


48,7% 


4071 thd 


52,1% 


3328 thd 


54,4% 


Average annua] growth '/% 






8.4% 


1984-1990 


6.6% 


1990-1993 


7.0% 
6.8% 


1993-1997 
1990-1997 


Interregional connections 


1479 thd 


33,1% 


2121 thd 


30,7% 


2291 thd 


29,3% 


2469 thd 


25,2% 








6.2% 


1984-1990 


2,6% 


1990-1993 


1.9% 
12% 


1993-1997 
1990-1997 


Total 


4466 thd 


100,0% 


6916 thd 


100,0% 


7008 thd 


100,0% 


9787 thd 


100,0% 








7.6% 


1984-1990 


4.1% 


1990-1993 


5.8% 

5.1% 


1993-1997 
1990-1997 



2) Scat capacity: Number of fcatJ on both the outgoing and the incoming (lights 
Source: OAG/ABC 



This conclusion is confirmed by the growth in transfer figures at several European airports 
collected by Kuehne (1999) for the last few years. (See appendix) 



Two indicators for hub developments in the EU 

The probability of increased hub&spoke connections as indicated by Tables 4 and 5 requires a 
more detailed analysis on the actual hub developments initiated by a limited number of 
European national carriers at their respective home bases. 
Two important indicators can be used to analyse these hub developments in more detail. 

• Increasing numbers of spokes as well as increased frequencies provided on these spokes 
should be reflected by a higher frequency growth of the home-based carrier when 
compared to the fi-equencies of other carriers at the airport involved. Therefore we first pay 
attention to this plausible increase of the frequency share as an indicator for the growing 
hub dominance during the period 1984-1997. 

• Furthermore an analysis of the daily waves pattern at the hub airport can reveal in more 
detail the actual type of hubbing. Numbers of waves (or connecting banks) as well as 
categories of connecting traffic (long haul and/or short haul) have to be taken into account. 

Hub dominance 



We analysed seventeen European airports mainly in the categories very large and large. (See 
also Tables 1 and 2). 



Table 6: Hub dominance 



Airports 


Home-based carrier 


Hub domtiuuice !990 
(in % tdUl 
frequeociei) 


Hub dominance 
199t(Tit% total 
jrequencies) 




category 1990 


category 1997 


London Heathrow 


British Airways 


n 


39 


^-1 


very large 


very large 


Paris Charles de Gaulle 


Air France 


39 


41 


*-2 


very large 


very large 


Frankiurt 


Luftharua 


4i 


% 


+Jl 


very large 


very large 


Rome 


Alitalia 


43 


62 


-MS 


very large 


very large 


Amsterdam Schlphol 


KLM 


3t 


43 


■t-14 


large 


very large 


London Gatwick 


British Airways 


19 


65 


+46 


large 


large 


Zurich 


Swissair 


50 


51 


+1 


large 


large 


Brussels 


Sabena 


4J 


47 


+4 


large 


large 


Paris Orly 


Air Inter 


24 


45 


^-24 


large 


large 


Munich 


Lufthansa 


50 


53 


+3 


large 


large 


Madrid 


Iberia 


59 


55 


•4 


large 


large 


Barcelona 


Iberia 


61 


51 


-to 


large 


large 


Copenhagen 


SAS 


47 


52 


■t-5 


large 


large 


Milan Linale 


Alitalia 


43 


M 


+12 


large 


large 


Vienna 


Austrian Airlines 


41 


34 


•7 


medium 


medium 


Milan Malpensa 


Alitalia 


29 


25 


-4 


medium 


medium 



Frequency shares for dominant carriers at major domestic hubs are usually higher (50-70%) 
than the U.S. gateways (less than 40%), where a larger share belongs to foreign carriers. Table 
6 indicates that most home-based carriers have consolidated their position at their home bases 
in Europe. Especially British Airways at London Gatwick, Air Inter at Paris Orly and Alitalia 
at Rome Fiumicino have substantially increased their frequency share.' One has to be careful to 
use these figures as a single indicator for hub developments. The example of British Airways at 
London Gatwick illustrates the dangers of misinterpretation. A co-ordinated wave structure is 
missing in 1997 despite an increase of the frequency share by 46 % in the period 1990-1997. 
(See diagram 1). 

Diagram]: Trafdc patterns of British Airways at London Gatwicic 



British Airways/London Gatwick 1990 



300 



g 100 
>; 50 



3 



■ 100 

■ 150 



H i . i m I M l m i M I mj i M m M M M M M m ^ 



6r«p«»vt>^jiti««» «n* *»««*$ 



1 1 1 1 ■ 1 1 1 H 1 1 1 i 1 1 1 1 1 1 i t H I 1 1 



rnYm i jY i . i VNV i V i 






t ':iy}\wft})Myi i^ '. 



ff^ t^i ^tp^f $ (k4fir^^ 



WWM. 



i:i^?<^i^<^j>^?!li^ 



200 -Liiiii 



79 99»999>993 

•-oirtf^^^ii 9=>ii 

— •- — — — 01 oj 

BEURdep BICAdep DEURarr OICAvr I Time Of Day 



British Airways/London Gatwick 1997 



t M** f l " lil»***'* " - H > H" l " l t M"f 1>'ttt " >'>H» f 




)At«ec0Ahmtatt>«paFliiM&antf Attxafy 



39* 9 999 9 9999 

iiiiTTTTT'^'y'!' 

— OW^r- 3 99 9 9 999 

O- — r> in r-ff' — o 
— — -^ — *- cy fu 



. BEURdep BlCAdep OEURarr DEURarr j Time Of Day 



The other way around, notorious hubbing carriers do not always demonstrate a substantial 

increase in their frequency share. This is mainly a consequence of our selectively focusing on 



" A HHI value, reflecting the frequency sluucs of the viinous airlines, is interuion:iil\ not used here, since tlie 
separate share of tlie doniinanl earner as siicli can no longer be recognised 



the national carrier alone, without taking account of the impacts of alliances and stakes in other 
carriers, which also operate at the home base of the alliance partner. For example, KLM alone 
shows a frequency share at Amsterdam Airport of 45%. However, if the frequencies of the 
alliance partners and subsidiaries are also included, the frequency share rises to 69%. Even a 
stagnant frequency share during the period ofl 990- 1997 can go hand in hand with a strong 
restructuring of the traffic pattern towards a hubbing system. Sabena, for example, was able to 
reorganise the daily frequency pattern as an Euro hubbing system without any substantial 
change in its frequency share at Brussels as diagram 2 demonstrates. 

Diagram 2: Traffic patterns of Sabeiui at Bnuaels airport Zaventem 



Sabena/Brussels 1990 



300 

250 

200 

S" 150 

% 100 

>. 50 

c 

a 

S" -50 

"" -100 

-ISO 

-200 



iiii i iiii iii iiii 



iii i iiiii i ii ni iiiiiiil i i n iiiiiii iir i ii iiii rmii iii ir ii 



"- ~ J SHN)))W»$%^$t|Mkl|^>»w|t^ 



n 'f'i'tif MHMM ti H '<ti MmnM t n i m > M »t v w m;n wwt K iiiiiiii n ii n»n t M '4 w <' 



i MMm M MMH IIIi mm * OHHM I MM 



' A * J . ' A ' J * '"JJ"J ' . . ".L ' " '* ^ * ." *** " ' " *** ■ *'*'**< 






., ..., . . > ., ,, ..,.. ^Jj , »4lli |U(i»i il i <»|ilii l 4^ ^ 

- :;^ ~f- . -Ic-'ift^'^l— Mow , .5 



)l*l* H iW«4tM»I H * H» i 



» t<<ll4<tWII HH I HmHm i H 4 Um W«*«W* 1 III H IIl 'i tl 



i i rmimi i rrr i m i iTi ii i iiiiii tr iiiiiii Ml ili r ilii m i n iiiiiiiiiiiii r i i iii r iii i lli r i t 



Of 



— *- *. ^ «- 01 Ol 



{■EURdep aiCAdep QEURarr QICAvr I Time of Day 



Sabena/Brussels 1997 



300 
250 
200 



^ ISO 

s 

% 100 

>» 50 

c 

U 



-100 
-150 
-200 



I [^^'"'i' ' I "^' ' II iiiiiiiiiiiii iiiiii 

H y iii j i MM II j n il I'l l ! Ill rti i nj Illlii rt i lm i XUt ii r iii 



n il II M IIIiliilililliliiiiiiliiiiiii M iiiit M iiiiiiii IIIIII iiiiiiiii mmf lll YiTf i if i f i T 



i'i>'i " i n l iit ii iM i i ii i iii i. 



i n iiiiii i i m ii u i n iiih^ n i 




l U llilli n lll iV ii mMi^ M il i nnnh iil hl li n l T ili nnnn ii n i n 



MtmeixiintiM D<(>artur«i mihtitMt 



^^^ 



OI**lA«0OM^lA40O0J^ 

— •■ — — — 01 01 



■ EURdep aiCAdep OEURarr OICAvr | Time Of Day 



Wave structures 

As indicated by Bootsma (1998) hub systems at an airport can be classified from the 
operational point of view by the triple (N,H,S), in which N= the number of waves, H= hub 
repeat cycle, and S = the stabling system (home based or not, or mixed). From a functional 
point of view the type of connection waves completes this triple: combined long-haul/short- 
haul waves or simple short-haul waves. 

The daily frequency patterns of the airports analysed indicate that in 1997 six out of seventeen 
airports demonstrated co-ordinated wave structures according to table 7. It is plausible that in 
the near future also Malpensa will rapidly change its position within table 7: the hub dominance 
of Alitalia strongly increases in 1999 after the opening of the renovated airport and Alitalia is 
now developing a wave system at this airport. 

The impact of liberalisation can be derived from the differences between Table 7 and Table 8. 
Before the liberalisation the hubbing phenomenon was non existent. Since the liberalisation 
however, hubbing has become a prerequisite for airiines to enable the new network 
competition. 



lahlc 7: Wave slnicliiri-s IV9" 



iiMiiiiiiiii 




^^m^S^^^^Si^Mimeuetixm 


Wimmaxmimmmimmmmm 


iiiii;li:iiiiiiiiiii 




"MiMnim^iif-MMWrnMi- 


limited 


'.i3ibseniW:M^Mmm'M 


liiiiiiiilli 

iiiiliiigiiiisii 
liiiiiiiiiiiiiiiil 


high 


Munich (4) 
Frankfurt (4) 


Zurich 
Copenhagen 


London Gatwick 
Madrid 
Barcelona 
Rome Fiumicino 
Milan Linate 


; Hab dmninance^ 

lilllllliiliiiili 


medium 


Bnuseb (3) 

Paris Charies de Gaulle (5) 

Amsterdam (3,5) 


Vienna 


London Heathrow 
Paris Oriy 




low 






Milan Malpensa 



Table 8: Wave stnictur 


■es 1990 

■K"-"'"'ffr,^i 




■: ■■■ " , '// 




Present 


United 


absent 




high 




Zurich 


Munich 
Madrid 
Barcelona 




medium 


Frankftirt 


Copenhagen 
Vienna 


London Heathrow 

Brussels 

Paris Charles de Gaulle 

Milan Linate 

Rome Fiumicino 




low 




Amsterdam 


London Gatwick 
Paris Orly 
Milan Malpensa 



In 1990, only Frankfurt airport demonstrated a clear daily peak pattern. Five other airports also 
showed a rudimentary wave structure. From this column only Amsterdam airport has been able 
to move to a fully-fledged hub airport. Paris Charles de Gaulle however, shows the largest 
change within a very short throughput time; from a non-hubbing airport to a complete 
hub&spoke system. 

All in all it can be concluded that hubbing is a clear phenomenon in Europe nowadays, be it at a 

slowly increasing number of airports; Munich, Paris Charles de Gaulle, Frankfurt Brussels and 

Amsterdam, and Copenhagen, Vienna arid Zurich in the second range. 

In the context of airline alliances it can be expected that more airports will develop a clear 

hubbing pattern in the near future. The geographical concentration of the currently operating 

hubs indicates a rapidly intensifying competition between the airline networks rooted at these 

hubs. 

Huh categories 

From a functional point of view hubbing can be fijrther categorised by looking at the mixture 

oflong-haul and short-haul operations. Following this approach, five hub types are relevant; 



the ICA galeway hub which connects short haul Euro destinations/origins and long haul 
ICA origins/destinations as well as a limited number of ICA-ICA transfers; 
the Euro hub which focuses on the connectivity between European origins and destinations; 
the Combi hub which integrates the ICA gateway function and the Euro hub function; 
the Random hub which provides a high random connectivity due to the large volume of 
arn\ ing and departing flights so that waiting time intervals remain within acceptable limits 



-Itib tiominancc of a home-based earner is strong if the frequenc\ share is larger than 50% and low if the 
.'cqiiciic\ share is smaller than 30%. 



in the perception of the consumer; and 
• the Non hub, which does not provide any useful connectivity to the transfer passenger. 

Table 9 contains the classification of a number of major European airports according to these 
five concepts. 

Table 9: hub-airport types 



HuJ)-atrpott 


,„...!?.y.\'...?.?.*.'?.8P.ry...^.?.?!?..... 


hub category 1997 


Munich 

tart«CH«fles($eG#uUe 

Bmiseb 

Am»tef(j4m 

'\^<nn« 


Combi-hub 

Non-hub 

Non-hub 

Non-hub 

Gateway 

Non-Hub 


Combi-hub 

Eurohub 

Combi-hub 

Eurohub 

Combi-hub 

Euro-hub 


London Heathrow 


Random hub 


Random hub 


London Galwicic 
Some fiumieino 
Zurich 


Non-hub 
Non-hub 
Gateway 


Non-hub 
Non-hub 
Gateway 



Concluding remarks 

The foregoing analysis indicates that the liberalisation of the EU airline industry went hand in 
hand with an ongoing transformation of national EU airports into different hub types. 
The assumption that congestion at the major European airports would be a sufficient reason to 
counter this development into a point to point network system is not sustained by the figures 
derived from the ABC/0 AG database, available at the Dutch Civil Aviation Department. 
On the contrary, the process towards a more sophisticated hubbing system in Europe is well 
under way. Not only more airports are getting involved in this process, also a hierarchy of hub 
airports within alliances may be plausible as a next step in this hubbing process. However, the 
volatility of cross border airline alliances dictates the relative (in)stability of these hub airport 
systems during the next few years. Whether the currently emerging multiple hub relationship 
will hold, is an unanswered question. For example, this question relates to Copenhagen and 
Frankfurt in the Star alliance, Milan Malpensa and Amsterdam in the Wings alliance, Zurich 
and Brussels in the Qualiflyer alliance and London Heathrow and Madrid in the One World 
alliance. 

Changes in airiine alliances can have tremendous impacts on airport planning in the EU during 
the next decade. If for example Air France is incorporated in the Wings alliance through its 
close connections with Continental (the carrier partly owned by KLM 's partner NorthWest), 
Paris Charles de Gaulle might become the primary European hub in an AF-AZ-KLM system. 
The near future probably clarifies whether the number of primary hubs in Europe will be 
limited to London, Paris and Frankfurt and followed by a number of secondary hubs like 
Munich, Copenhagen and Brussels The ahernative would be a more decentralised multiple hub 
system will emerge due to increasing congestion problems at these primary hubs. 
Finally, with regard to the European airport system we can conclude that the crucial question is 
not ftubbing versus hub-bypassing" but 'Single layer hubbing versus multi layer hubbing". 
Airline co-operation and alliances will ultimately determine the answer on this question. 



References 



Bootsma, P.D., (1997) Airline Flight Schedule Development -Analysis and design tools for 
European hinterland hubs, PhD thesis Twente University. 

Tretheway, M.W., and Tae H. Oum, (1992), Airline Economics: Foundations for strategy and 
Policy, University of British Columbia, Vancouver. 

Kuehne, M. (1999) Transfer figures at twelve European airports, DFV Stuttgart. (See 
appendix) 



Environmental sustainability, airport capacity and European air 
transport liberalization: irreconcilable goals? 

Brian Graham 
School of Environmental Studies, University of Ulster, Cole ra inc. Northern 

Ireland BT52 ISA, UK 



Introduction 

The general aim of this paper is to discuss the principal tensions that exist between 
policies for air transport liberalization in the European Union (EU) and those directed at 
environmental sustainability, conflicts which come together in the nexus of airport 
capacity. While concentrating on one particular transport mode, the discussion is more 
widely informed by the mounting recognition that present and projected trends in 
mobility in Europe cannot be sustained and that, more generally, 'the belief in the 
desirability of perpetual growth in mobility and transport has started to fade' (Greene 
and Wegener, 1997, p. 177). As Black (1998, p. ) argues, 'history would suggest 
that it is not the transport vehicle... .[but its]. ...excessive use... .that creates the 
problem'. 

At least four interested parties or stakeholders can be identified in the 
relationships between environmental sustainability, airport capacity and European air 
transport liberalization. The airlines themselves, transformed by liberalization into a 
resolutely free-market industiy, can often express unreconstmcted attitudes to 
environmental issues, which are perceived to interfere in their primary goal of making 
money. Secondly, environmental objections originate in the concerns of wider society, 
although these may range from empirically verifiable complaints about air transport 
noise and atmospheric pollution to the actions of an idealistic lobby prepared to sacrifice 
economic growth to its perceptions of environmental needs. Demands for 
environmental quality increase with standard of living (Maddison, 1996), and it is the 
experience of airport operators that the maximum number of complaints regarding 
aviation originate from high-income residents in their immediate hinterlands. 
Ironically, the demand for air transport also increases with income and those members 
of society complaining most vociferously may also be those flying most frequently. As 
airline customers, they want maximum mobility, combined with cost or status 
advantages. Thirdly, if it is accepted that unconstrained mobility is no longer a feasible 
goal for society, then regulators are required to somehow ration demand for airport 
capacity and reduce the environmental externalities of air transport. Finally, the airport 
operators occupy the interface between this conflicting mesh of interests. 

The essential assumption that underpins the paper's argument is that - at an 
aggregate level - the EU lacks sufficient airport capacity - however defined - to 
accommodate projected growth trends in air transport, and that the provision of 
extensive additional infrastructure is extremely unlikely because of environmental 
constraints. More specifically, the paper has three objectives. Initially, we address the 
concept of environmental sustainability and its relationship to capacity issues in EU air 
transport. Secondly, the problems of European aiipon capacity are assessed, as is the 
potential for modal shift. Finally, the bulk of the discussion is given over to the ways 
in which the often incompatible interests and goals of the various stakeholders outlined 
above define complex tensions that immensely complicate any rcsokition of the 
relationships between environmental sustainability, aiiport capacity and liberalization. 

The concept of environmental sustainability and its relationship to 
capacity issues in European air transport 

Transport in general constitutes the most important negative environmental externality 
of the Single European Market (SEND, creating noise, atmospheric pollution and 
consuming large areas of land, while being dependoii; on non-renewable energy 



resources. Although its aggregate impact is minor compared to road traffic, air 
transport accoimts around 10% of all transport energy consumption in the EU and is 
responsible for approximately 15% of all COi emissions (Stunners and Bourdeau, 
1995). However, the technological returns on reducing air transport's negative 
environmental externalities are diminishing so the sector's veiy growth seems likely to 
ensure that this impact will increa.se in the future. In addressing the relationships 
between air transport infrastructural provision and the environment, two key terms - 
sustainability and capacity - require definition. 

SusUiiimbiUty 

The meaning of sustainability to transport has occasioned widespread discussions in 
recent years, not least because it is a qualitative rather than operational term (see, for 
example, Pearce, 1993; Black, 1996; Nijkamp and van Geenhuizen, 1997). The 
common thread in these debates is provided by the dual invocation of sustainability put 
forward in the 1992 Rio Declaration, which attempted to reconcile the needs, especially 
those of the world's poor, with protecting the environment's capacity to meet present 
and future needs. Thus Black (1996, p. 151) defines sustainable transport as 
'satisfying current transport and mobility needs without compromising the ability of 
future generations to meet these needs'. According to the Aviation Environment 
Federation (1997), sustainability describes integrated transport systems and 
infrastructure, which enable the socio-economic needs for movement of goods and 
people to be met within the long-teiTn carrying capacity of the planet's ecological 
systems. Greene and Wegener (1997) argue that sustainability as applied to transport 
has three basic conditions: that: the rates of use of renewable resources do not exceed 
their rates of generation; the rates of use of non-renewable resources do not exceed the 
rate at which sustainable renewable substitutes are developed; the rates of pollution 
emission do not exceed the assimilative capacity of the environment. Air transport fails 
outright to satisfy the first two conditions and probably also the third. In the longer 
term (perhaps 2050+), global air transport is not sustainable on any basis because there 
is, as yet, no feasible substitute fuel for oil, hydrogen-based fuels being the only 
apparent possibility. 

It must be emphasized that in addition to concerns with environmental carrying 
capacity, sustainability also invokes connotations of social needs and equity. The 
problem is that tactics aimed at achieving social equity also encourage mobility. 
'Modern transport affords mobility, facilitates post-Fordist production and allows 
political cohesion. Degrees of access to transport networks affect social patterns at all 
levels of spatial aggregation' (Button and Nijkamp, 1997, p. 215). The equity 
implications of mobility creation are central to the social economy model that underpins 
the ideological constmction of the EU and its concern that geographical location should 
not be the primary determinant of the life chances of the Union's 370 million 
population. The European Commission in general tends to refer to 'sustainable 
mobility', which is unfortunate as there are strong grounds for believing that mobility 
as currently practiced in developed countries is itself unsustainable (Fergusson et al., 
1994); infinite mobility is not infinitely desirable (Bleijenberg, 1995). It is access, not 
mobility per .vf , which is the critical issue in social needs and the enhancement of 
accessibility is a key proce.ss in Commission policies aimed at alleviating regional 
disparities in wealth in the EU. An efficient transport system is also vital to the 
integration and efficient functioning of the SEM. Inevitably, however, the provision of 
transport infrastructure aimed at these goal also encourages mobility. In sustainability 
terms, therefore, the EU requires a transport strategy, which reconciles a curb on 
mobilitv with competing demands for accessibility related to: the nC'id for competitive 
elficiencv: the EU commitment to geographical accessibility and social equity for all its 
citizens; and environmentally sustainable de\'elopment (Button and Niikaiiip. 1997). 

A major difficulty in achie\lng any such resolution. lio\vc\ ei , i^ that the 
responsibility for polic_\ -making u ithin the EU is dn ided between the various 
Diieciorates-Gencral o\' the Commission, no less thim four - DCs W . VII. XI and XVI 
(dcaluig rc.spccti\el\ uilh competition, transport. cn\ ironmont and regional 
de\clo|)iiicnl) - being directly involved in i^^ucs related K^ air iianspoii and tht 



ic 



environment. Already at odds over the regulation of competition within EU air 
transport, DGs IV and VII are concerned primarily with the market etYiciency of the 
industiy and the implementation of the Single Aviation Market, effectively created by 
the Three Packages of airline liberalization measures, introduced progressively between 
1988 and 1997. This policy initiative, which originated from DGVII, is concerned 
directly with promoting competition in air transport and removing barriers to market 
entry. However, as is a characteristic of all transport modes, such policies do not 
encourage individual restraint on the part of any one airline, because such actions 
would not be 'compatible with rational self-interest, not least while any other 
[company] reserves the right to use the resource [airport capacity] as much as they 
choose' (Maddison, 1996, p. 10). DGIV clearly regards airport capacity as more than 
a straightforward resource. It seems intent, for example, on using regulation of 
mnway slots - the most obvious and contentious manifestation of'capacity - to promote 
intra-EU market entry, particularly by low-cost airlines. The slot is a strategic weapon 
in a competitive market-place, the major European airlines having a vested interest in 
ensuring shortages at their principal hubs (so long as they themselves have sufficient) 
in order to deter market entry and control competition. The Association of European 
Airlines (AEA) estimates that a ninway is at saturation point if 70% of its slots are 
being used; peak-time slots would have been used fully long before that. In attempting 
to regulate the anti-competitive connotations of the consolidation of EU airlines into 
internal alliances and more extensive global coalitions, DGIV opposes the concept of 
airline ownership of - and trade in - slots and is demanding that British Airways (BA) 
and Lufthansa surrender significant numbers at Heathrow and Frankfurt, respectively, 
in return for regulatory approval of their separate global alliances. 

More widely, DGVII is responsible for the Common Transport Policy and its 
principal modus operandi, the multi-modal Trans-European Transport Network 
(TETN). Its role is to enhance accessibility and integration, while harmonizing national 
networks into a macro-network for the EU as a whole, not least by providing missin<> 
connections (often at border locations) and the attempted elimination of bottlenecks 
(CEC. 1994; Banister et al., 1995). While the TETN focuses on High Speed Trains 
(HSTs) rather then air transport for inter-city public transport within'the EU, its 
commitment to competitive efficiency also includes inter-modal complementarity. Thus 
an essential element of the netwoik lies in the development of the most important EU 
airports as multi-modal high-speed interchanges. 

The TETN is also linked to other EU policies and objectives being articulated by 
DGXVl through the Regional Development and Structural Funds, and aimed at 
operational izing the commitments to social solidarity, cohesion and convergence that lie 
it the heart of European integration. In particular, this requires investmentin transport 
links to rural and peripheral areas, the assumption being that long-term cohesion- 
oriented policies demand a coherent and efficient transport system guaranteeing 
continuity of .service (CEC, 1996a). In DGXVI's terms, the notion of 'sustainable 
mobility' (not only in the sense of eniissions and noise but al.so of the social equity 
connotations that underpin the integrated spatial planning ethos of the TETN) has 
become the 'central goal of transport policy' (CEC. 1996a, p. 76). In reality, however, 
convergence and cohesion policy may simply ensure the construction of transport 
infrastructure that otherwise would not have been built, under-utilization of expensive 
resources providing another dimension to the aiiport capacity debate. 

It is required of EU transport and cohesion policies that they be environmentally 
sustainable, but it can be argued that both enhance the demand for mobility (without 
necessarily improving accessibility), whereas environmental policy - the remit of DGXI 
- tends to assume that present and projected demand for mobility is unsustainable and 
must therefore be reduced. In the EU's Fifth Environmental .Action Programme, 
endorsed in 1993 and subtitled. 'Towards Sustainabilit\'. iranspon is identified as one 
i'!' live taigei sectors in recogniiion of the point that it can iiexer be cm ironmentally 
iiOLii!-,il. The Proizramme argues thai present trends in air lanJ roaJ) transport are 
IcuLliiig towards greater cn\iroiiincnlal costs - congestion, polluiioii. wastage of time 
aiui \aluc. damaize to health, and diuiifcr lo life I CEC. I9y6hi. 



The most recent assessment of the Programme, and Agenda 21, the general and 
politically compromised strategy for sustainable development set up after the 1992 
Earth Summit (and reaffirmed at Kyoto in 1998), concludes that the transport sector is 
displaying an increased awareness of the unsustainability of present trends (CEC, 
1997a). Traffic growth, however, is eroding attempts to move towards a sustainable 
system, air transport having a higher growth rate than any other transport mode. One 
stark conclusion is apparent; transport policy must be designed to reduce demand for 
mobility, a demand which is derived and can therefore be altered. But as Greene and 
Wegener warn (1997, p. 180), transport demand policies to mitigate the environmental 
impacts of transport 'are frequently dwarfed by countervailing market developments'. 
Nowhere is this more apparent than in the EU's air transport industry. 

Airport capacity and its environmental context 

Aiiport capacity takes several forms. It includes: airspace and the role of Air Traffic 
Control (ATC) techniques in the maximization of air transport movements (ATMs): 
ail-port infrastructure - runways, aprons, piers, and terminals: and terminals for 
terrestrial transport as airports (particularly the most important) are multi-modal 
interchanges. Above all, however, it is es.sentially the case in the EU that airport 
capacity is - or is soon to become - environmental capacity, with environmental criteria, 
rather than those related directly to physical infrastmcture capacity increasingly 
determining the magnitude of ATMs. Airports are increasingly left free to plan 
operations, provided that the sum total of the environmental impacts of their activities 
do not exceed a pre-detemiined level. 

Environmental capacity invokes a wide range of concerns, which include: noise 
from aircraft and surface transport; atmospheric emissions from aircraft engines; 
surface access congestion at airports; land-use severance effects of aiiports and their 
impact on visual amenity; effluents; and waste management. Noise in particular 
remains critical to environmental capacity because it is the principal source of 
complaints and the most likely cause of political involvement in restricting the use of 
existing - and further development of- infrastnicture. Commercial jet transports, 
currently in operation, are divided into Stage II and III types, classifications that relate 
to Chapters 2 and 3 of Annexe 16 to the Chicago Convention. All new aircraft must 
meet Chapter 3 requirements, although these were laid down as long ago as 1976. In 
1990, the International Civil Aviation Organization (ICAO) agreed to pha.se out all 
Chapter 2 aircraft. If these are to remain in service beyond 2002, the EU target date for 
final Chapter 3 compliance, they will either have to be hush-kitted to those standards, 
or re-engined. 

The enforcement of these requirements will not, however, remove the problem 
of aircraft noise. New aircraft are quiet, only when compared to their predecessors. 
Although the spatial extent of noise footprints around aiiports has been reduced, the 
problem of aural pollution will remain. The high-by-pass turbo-fan engines used in 
modern aircraft are probably already as quiet as is technically feasible although the 
possibility exists that, assuming the necessaiy investment in research, further gains 
might accrue from the development of prop-fans. More immediate reductions will most 
likely accrue from reductions in airframe noise, which may account for around 50% of 
total aircraft noise on airport approaches. Serious concerns exist, however, that 
increased ATMs are compromising noise reductions, while larger aircraft create more 
noise, even if they do comply with Chapter 3 limits. In these contexts, any 
enhancements to airport capacity - whatever their form - depend on a proactive 
environmental policy on the pan of airport operators, addressing not only noise but the 
entire suite of environmental externalities engendered by the air transport industiy. 
Although expenditure on environmental issues may not be justified in terms of direct 
economic costs and benefits, any future enhancement of capacity is predicated on a 
xisililc and efi'ectixe cn\ ironmen[al policy. 



The prolileni of European airport capacity 

The airport capacity prohleni 

European aiiport.s have long been perceived as having capacity problems although these 
have been offset by innovations in air traffic management and control. In reality, the 
picture is rendered more complex by three important factors, which, in turn, create a 
geography of airport capacity restrictions. These are: variations in the form'of airport 
infrastructure itself; the growth in demand for air transport; and the distribution of that 
demand for air transport. 

First, as Table 1 shows, runway capacity is a function of an airport's layout, 
parallel runways unsurprisingly supporting larger capacities than converging or single 
amways. The data also indicate that irrespective of geometi7, only marginal incresSes 
in runway capacity can now be achieved without the constaiction of additional 
infrastructure. There remains some potential in EU harmonization of ATC, and in 
innovations such as 'mixed mode' runway operations in which the same runway is 
used for landings and take-offs. 

Secondly, aggregate demand for air transport remains driven by GDP growth, 
although changes in industrial organization (especially just-in-time deliveiy) and 
lifestyle (particularly enhanced consumption of holidays, a market driven by increased 
real incomes) also contribute significantly. In addition, European air transport 
liberalization has helped grow the air transport market through price competition 
(Graham, 1997a; CEC, 1997b), as have the strategic actions of the major airlines in this 
reformed market-place - a point to which we return later. Growth rates for air transport 
in Europe have been rising consistently since the global slump in air travel induced by 
the Gulf War in 1991. In 1997, for example, AEA airlines carried 164.4m passengers 
on international routes, a 10.7% increase over 1996 (AEA, 1998). Future projections 
vary but the 'predict and provide' scenarios of the aircraft manufacturers offer some 
(rare) agreement. Airbus Industrie estimates an average annual growth rate in traffic of 
5.3% up to 2001, and a more conservative 4.6% betvveen 2002-201 1. Boeing (1998) 
is predicting an average 5% growth in air travel over the next ten years. Because such 
growth is exponential, these annual increments are equivalent to a doubling in demand 
every 12 years. Airbus estimates that the global population of passenger aircraft will 
double from 9,700 in 1998 to 17,900 in 2018, flights increasing by 88% over the same 
period. The number of runway movements is expected to increase' at rather lesser rates 
(Table 2), an imbalance which implies that at least some traffic growth is - and will be - 
accommodated by larger aircraft. One major imponderable in such predictions is the 
future impact of information technology on the demand for air travel. 

Thirdly, both the demand for air transport in the EU and congestion are spatially 
concentrated. Demand is heavily bia.sed towards Europe's most dynamic and urbanized 
vital axis, stretching from Manchester in the north-west and Helsinki in the north-east 
to Rome, Barcelona and Madrid in the south. This axis contains virtually all the EU 
regions with above-average GDP/capita and the most important airport hub systems 
(Graham, 1998). Airport capacity problems - and the congestion and pollution created 
by terrestrial transport modes - are also concentrated in this central vital axis, although a 
secondary nucleus comprises certain of the leisure-oriented aiiports of southern 
Europe. In essence, Europe is running out of airport capacity - however defined - in 
the regions in which demand for air travel is most hea\'ily concentrated. Although more 
than 450 European airports receive scheduled service, the 20 busiest - largely 
concentrated in the EU's dominant axis - account for about 55% of all scheduled seats 
and virtually all long-haul traffic (Fif^urc I ) (Boeing. 1998). While capacities have 
increased markedly at some of these airports, a now slightly dated survey of the 29 
European airports handling more than 5 million passengers per annum in 1994, 
estimateLl that by 2005. 25 will have runway, and 26 terminal, capacity shortages 
iespecti\ely f AE.A. 1996). All will be congested by 2010. While the maximum hourly 
movements at most major airports will increase h\ 2015. the extra capacit\' is 
in-.urficient to meet projected growth in alniu>i cvoiy iiNianco {Tcihic 2 i. Conversely, 
a substantial number of ailpo|■l^ capable of handling more than immediateU' local or 
regional traffic have adequate capacii\ lor the rore>eeabie t'liiuie {Fi'^iirc 2 i. and while 
many of these ser\e either secondaiy cities or ihc iiioic peripheral rejiion^ o\' the present 



and future EU, some are actually located within the vital axis, thereby providing some 
limited potential for traffic diversion. 

Coping with growth 

In comparison with the projected growth rates for air transport, the plans for 
constaicting new airport infrastmcture in the EU are modest. Munich Franz Josef 
Strauss - the last major greenfield airport to be built in Europe - and the reconstructed 
facilities at Milan Malpen.sa and Oslo Gardermoen, both completed in late 1998, largely 
replaced existing capacity, although obviously they also added some. The same is true 
of those aiiports currently in the planning stage - Berlin, Lisbon and Athens Spata. 
Terminal capacity is being increased at a number of aiiports although many of these are 
not capacity restricted anyway. While it is easier to get permission for terminals than 
runways, by late 1998 the planning inquiry into Heathrow's proposed Terminal 5 (T5) 
had sat longer than any other inquiry in UK planning history; even if approved, the 
terminal will not be fully operational until 2015-16. The construction of iiinways, or 
even their lengthening, creates even more strenuous opposition. Consequently, one 
recent survey listed 47 existing European airports at which terminal expansion is 
projected or in progress, but could identify only 12 instances of new runways being 
planned (Simon, 1998). 

Thus it appears that an irreconcilable tensions exists between projected growth 
figures for air transport in Europe and the provision of the infrastmcture necessary to 
cope with that growth. To put it more simply, the projected growth rates cannot be 
sustained within current or projected air transport infrastmcture capacity. It also seems 
fair to assume that the lack of political will to build additional capacity - either at the EU 
or Member State scale - owes much to the environmental opposition that such plans 
encounter. Consequently, policy initiatives to cope with the growth in air transport - 
most notably the TETN - have been directed as much at modal shift as at building new 
airport infrastructure. Some short-haul air traffic could be diverted to HSTs, which 
consume much less energy per pas.senger km, allowing airlines to concentrate on their 
unchallenged hegemony in intercontinental travel. For point-to-point business traffic, 
HSTs can compete effectively with air transport on inter-city journeys of less than three 
hours (approximately 500 km); the threshold extends to 1,000 km for leisure traffic. 
One leading EU regional airline, the German carrier, Eurowings, has admitted that 
regional air services are no longer worth flying if the journey time by rail is less than 
three hours {Fliglit International, 1997). 

The potential for HST modal shift was first demonstrated by France's TGV, 
which reportedly captured as much as 90% of the Paris-Lyon market. Elsewhere, the 
AVE .service has over 80% of the Madrid-Seville market, compared to the 33% share 
held by conventional rail in 1991 (CAA, 1998). In Germany, Lufthansa, which 
already uses Inter-City Express (ICE) trains, wants to shut down domestic air services 
from Frankfurt to Cologne, Diisseldorf and Stuttgart, but is facing difficulties in 
guaranteeing passengers the equivalent level of service). Although early HST 
development concentrated on city centre-city centre linkages, the most recent network 
additions exploit the added value offered by this mode when it interfaces with other 
high-speed systems (Thompson, 1995). The construction of TGK stations at Paris 
Charles de Gaulle and Lyon Satolas originated the process now being pursued through 
Ihe TETN, in which the integration of road, conventional rail, HST and air transport 
modes al major airports will produce a succession of sophisticated mainports across 
Europe, allowing the seamless integration of intra-urban. regional, national, 
international and global traffic flows (Graham, 1995). These will include Brussels 
National. Amsterdam Schiphol. Diisseldorf Rhein-Ruhr. Munich Franz Josef Strauss, 
Frankfurt International and Milan Malpensa, although the TETN will not be completed 
until after 2010. 

Despite the poiciuial for modal shil't and the increasing iniceraiion of airline and 
l-IST operations land e\en ownership! m the EU. ihis is not in itself ;i comprehensive 
soluiion to the problem of air transport capacity, partly because of the segmented nature 
oliiic air u-ansport market. Scheduled passenger scrviccN - which [tlace the greatest 
demand on airpon capacit>- - accoum foi' only half the pa>seiigcr inarkci. The potential 



for HST modal shift is largely restricted to the EU's dominant axis where the dense, 
juxtaposed city-pair markets necessary to support investment are concentrated. The' 
strategy is also irrelevant to the Inclusive Tour (IT) industry, which accounts for the 
other half of the passenger market. Arguably, however, this sector is - inadvertently - 
more compliant with sustainability requirements, its economics strongly encouraging' 
the employment of the most modern fuel-efficient aircraft types at very high load° 
factors, and often utilizing off-peak times or under-utilized (or capacity-rich) regional 
airports. The freight market is distinctly problematical, however, given its dependence 
on night flights and older aircraft (even if hush-kitted); also, much intra-European 'air 
freight' is actually tiiicked, thereby adding to road congestion and pollution. In sum, 
therefore, modal shift offers some contribution towards alleviating problems of airport 
capacity but it is not remotely a comprehensive solution. 

The stakeholders in the environmental sustainability- airport capacity 
relationship 

Tlie regulators 

The incompatibility of growth trends and projected infrastructure availability, combined 
with the limited potential for modal shift, suggests that any resolution of the tension, 
which exists between projected demand for air transport and airport capacity, lies in 
regulatory measures to curb that growth altogether. It has long been accepted that 
airport operations can be restricted for environmental reasons, night curfews or quotas, 
and bans or restrictions on Chapter 2 aircraft being obvious examples. Again, political' 
factors can influence capacity limits, one notable example being the legally-binding 
agreement preventing the construction or opening of a second runway'at London ° 
Gatwick (the busiest single-runway airport in Europe) before 2019. The slot-capacity 
of the existing infrastructure at Diisseldorf Rhein-Ruhr, for example, is constrained for 
environmental reasons, while the Dutch government's attempt - albeit now revised - to 
place growth limits on Amsterdam Schiphol are, most probably, a precursor of more 
widely applied and increasingly rigorous interpretations of the meaning of 
environmental capacity. 

The execution of environmental policy is critically dependent on regulatory 
intervention impacting both on demand for, and supply of, transport. Evidence 
suggests that the implementation of environmental policy is driven by threat, 'the 
dominant influence on a company's investment in environmental technology [being] the 
need to comply with regulations' (Hitchens, 1997, p. 816). The tenor of such 
regulation has changed, however, from a command/control to fiscal basis, which 
assumes that demand - whatever the mode - is exaggerated because transport does not 
meet its real costs (Stanners and Bourdeau, 1995). While not yet widely applied, 
European Commission policy on the environmental repercussions is clearly expressed, 
the 'polluter pays' principle being the common thread linking its various environmental 
regulations (CEC, 1997a; Hitchens, 1997). In other words." fiscal instruments should 
be employed in ensuring that transport users pay the full costs of their actions, the 
objective being one 'of indirectly influencing the supply of transport or the demand for 
it' by market mechanisms (Button, 1994, p. 128). The Aviation Environment 
Fedei-ation (1995), for example, advocates pricing discrimination in favour of more 
efficient and larger long-haul aircraft, a strategy which would optimize airport capacity 
w hile exploiting air transport's incomparable advantages for intercontinental travel. 
Again, the Commission is considering a kercsene tax on aircraft fuel. 

Any market-oriented initiative to address the interaction between aiiport capacity 
and environmental issues in Europe is rendered more complex, however, by the ways 
in which policies and legislation emanate from - and interact at - a variety of scales and 
agencies, ranging from globally-binding agreements, through the Commission and 
individual Member State governments, down to the micro-le\el of an individual aiiport 
and its local planning authority {Fi;^ure .?). .N'oise. for example. in li global issue 
experienced at the scale of the immediate airport localit}-. al.so the scale at w hich 
cuniplainis about the eflects of aircraft engine emissions on heahh are most frequently 
expressed. .Al various slaee> in this regulatory hierarehw houcver. iheie are missing 
Niagcs. which, in turn, open up inerea>.ed opportunities for unilateral action at the 



supranational, national or local scales. For example, the failure of ICAO to agree 
international post-2002 noise limits, when Chapter 2 aircraft will finally be bimned at 
EU airports, has encouraged the Commission, Member State governments and 
individual airport operators to introduce their own noise rules and surcharges in 
reaction to more localized pressures. Thus the Commission's consultation paper, Air 
Transport and the Environment (CEC, 1998), states bluntly that further improvements 
are required on noise and emissions to ensure the sustainable development of air 
transport. 

Such regional initiatives are inherently unfair to the airlines involved, subjecting 
them to penalties which do not necessarily apply to their global competitors. Moreover, 
their effect is compounded by individual local restrictions, for example on Chapter 2 
aircraft, the incidence of which, in turn, may reflect factors such as the fleet 
composition of an airport's major users. German airports tended to introduce stiff 
penalties on older aircraft once Lufthansa had a Chapter 3 compliant fleet, whereas 
Dublin, for example, continues to suffer Ryanair's hush-kitted Chapter 2 Boeing 737- 
200 fleet. Freight airlines in particular have been targeted by locally devised noise 
restrictions, as for example at Nuremberg and Liege (Flight International, 1998), partly 
becau.se they fly almost exclusively at night, often with older, hush-kitted aircraft that 
are only marginally Chapter 3. Again, the lack of binding global agreements , and - 
despite a lot of research - a failure to fully understand the effects of high-altitude 
emissions and contrails on global warming and ozone depletion, may culminate in 
unilateral action on Nitrogen Oxide (NOx) limits. In addition, and in common with 
several Member State governments (including the UK), the Commission lacks an 
integrated environmental management policy with regard to transport in general. Nor - 
excepting DGIV's attempts to regulate slots and their contested ownership - has it 
formulated any strategy to deal with airport capacity constraints, the principle of 
subsidiarity, that competence be exercized at the lowest level - as near to citizens as 
possible, thereby maximizing flexibility and local discretion, apparently applying to 
airport capacity issues. 

The airport operators ami society at large 

This mesh of different scales and the gaps in the regulatoiy hierarchy, combined with 
the ineffectual nature of some legislation, the array of motives involved, and the 
absence of centralized policies (the TETN excepted), which might reconcile the demand 
for air transport, its capacity constraints and environmental concerns, ensure that 
decisions on aiiport capacity are often made at the local level in agreements between an 
individual aiiport operator and its immediate planning authority. It is generally the case 
that planning permission for capacity increases depends on operators providing an 
integrated, locally-acceptable resolution to the entire suite of environmental externalities 
associated with air transport. Compliance with environmental regulations alone is not a 
sufficient strategy for an aiiport operating company. It has to design and implement a 
proactive policy that addresses a raft of environmental concerns (Figure 4). These 
include: 
• effective monitoring and regulation of aircraft noise on the ground and in the air; 

fi.scal penalties on noise offenders and best practice instruction for habitual 

malefactoi's: 

night curfews or quotas: 

public transport surface access targets: 

monitoring of airside and landside emissions: 

the reduction of energy consumption in terminals and by airport vehicles: 

the recycling of aiiport waste: 

monitoring water quality and reducing the impact of coiuaiiunaiits - particularly de- 
icing fluid and oil - on groundwater: 

niCLisures lo limit the \isual impact of an aiipori and ii-> land-ii>c vc\eiance effects. 
liic\iiab!\ , because the airport capaciiy-environmoiital ichmoii is m) often 
incLliated at the local scale, noise tends to be the predominant source of complaint from 
eoninumiiies in airport hiiiierlands. Ultimately, e\en the largest imeivontinental hub is 
mteraciim: w iih a local cominunitv and tiie concerns of us inhabiuuits. a ne.\us of 



conflict of interests dominated by environmental issues, primarily noise. In 1997, the 
Dutch government proposed a cap on movements at Amsterdam Schiphol, which ' 
ranked fourth in Europe and twentieth in the world in 1996 for passenger traffic. The 
airport, which has a theoretical capacity of around 650,000 slots per annum was 
restricted to only 360,000 slots for 1998, compared to the 400,000 requested by the 
airlines. At the time of writing, it has just been announced that these limits are to be 
revised, the airport's capacity being allowed to grow in annual increments of 20,000 
slots to a ceiling of 600,000. Environmental protests are likely to follow this decision 
even if future runway developments are designed to minimise noise externalities. 
Schiphol will maintain its tight controls on night flights and al.so continue to operate its 
noise 'budget' in which aircraft are given values according to the time of day and type 
of aircraft (Cameron, 1998). The aiiport is actively discouraging Chapter 2 aircraft 
which 'cost' too much, while giving financial bonuses to the quietest aircraft. In this 
context, it is important to remember that not all aircraft qualifying under Chapter 3 
comply equally with those standards. The principal opposition to T5 at Heathrow (the 
world's most important international aiiport) is from the surrounding local authorities, 
who - not unsurprisingly - question the figures put forward by the operator, BAA, and 
the principal airline user, BA, that, due to the use of larger aircraft, an additional 30 
million passengers per annum could be accommodated through a marginal increase in 
movements. 

Local authorities, however, have no direct control over the negotiation of noise 
standards, which are effectively global agreements (although they are^concerned with 
their effective implementation and monitoring), and thus m"ay be exercized more directly 
by other manifestations of the adverse impact of airports on their immediate 
environments. Chief among these is the issue of surface access, and the contribution 
niade by airports to road traffic congestion and pollution. European airpon operators 
increasingly recognize the importance of modal shift to public transport, not only for 
pa.s,sengers who may use the aiiport only infrequently, but also for employees who 
travel to and from it on a daily basis. All the major UK airports, for example, have 
ambitious public transport access targets. Heathrow is aiming at 50% for all journeys 
(compared to the present 34%), Gatwick has a 40% target for passengers by 2000 
(now 31%), while Manchester is seeking to increase its current 15% to 257c of all 
journeys by 2005. 

Because capacity - however defined - is related so intimately to local concerns, 
an aiiport business can grow, only if it minimizes the impact of its expanding activities 
on that environment and its residents. The circular problem for the aiiport operator is 
that having developed and implemented an environmental policy in order (possibly) to 
be allowed to expand capacity, the externalities of the resultant growth in air traffic 
created by that additional capacity may outstrip the benefits of the environmental policy. 
Consequently, airports have had to develop effective and continuous methods of 
communicating with those residents, who share Western expectations of an enhanced 
quality of life. Local protests are also often confiated by concerns over property values 
in urban areas adjacent to airports. One study of the vicinity of Manchester concluded 
that noise effects on residential property values could not be separated from a wide 
spectrum of neighbourhood and environmental variables infiuencing property values. 
Although hou.se prices were lower in the noise-affected areas, these^properties would 
still have commanded lower prices, even if they had not been located under the fiiaht- 
palh (Pennington et al., 1990). Subsequently, 'the .N4anchester data was re-worked by 
Collins and E\ans (1994) who did find a relatively minor noise component in house 
values. 

Tlic airlines 

While the adoption of a dynamic environmental strategy is clearly a rational decision for 
EU land cnhcri airport operators, and is perhaps the most important factor driving their 
businesses. c\cii if it does not resuh in an_\ additional capacity, it is rcadil_\ apparent 
ihai another sot of tensions exist heiweon airports and iheir principal customers, the 
airlines. |-oi' ilie latter, the definiiioii of a rational business straieiiv wiiliin the context 
ol European liberalization includes pracuccs thai exacerbate the alivadv hea\ y pressures 



10 



on airport capacity, especially at the largest hub airports on which the major carriers' 
networks are centred. For some airlines, the preferred environmental policy is 
probably not to have one; at best, a company will develop an environmenta Istrategy 
only if it is beneficial in profit terms to do so. 

The current actions of European airlines in response to the liberalized aviation 
market-place are largely incompatible with the precepts of environmental sustainability, 
and (he likelihood that any resolution of the capacity-growth tension can be achieved 
without curbs on demand. Four factors can be identified, all - possibly excepting the 
last - impacting negatively on capacity: 

• the development of hub-and-spoke systems; 

the dependence on increased frequency of service as the primary strategy in 
accommodating growth and also its role as the principal competitive weapon; 
the growth of low-cost airlines; 

• the development of alliances and code-sharing. 

1 . Hub-and-spoke systems 

The route networks of the largest EU airlines are being reconstructed from radial point- 
to-point to hub-and-spoke systems. The latter involves a dominant carrier operating 
synchronized banks - or waves - of flights in which the hub-arrival times of aircraft, 
originaiing from cities at the ends of numerous spokes, are co-ordinated into a short 
time period. After the minimum interval necessary to redistribute passengers and 
baggage, an equally large number of aircraft departs to the spoke cities. This pattern, 
which is repeated several times during the day, is essentially a supplier-driven strategy, 
maximizing the on-line (same carrier or alliance) connections available to a particular 
airline at the hub airport (Dennis, 1994; Graham, 1995). Hub dominance is a large 
incumbent's most effective defensive tactic in a liberalized market because, especially 
when combined with airport congestion and linked to an alliance strategy, it offers the 
real possibility of pre-empting - or at least controlling - competition at a particular 
aiiport. Its efficient operation is dependent upon available runway and terminal 
capacity to handle the peaks, combined with extensive feeder connections, often 
employing smaller aircraft operated by regional airlines. 

The cumulative effect of EU hub-and-spoke operations is to concentrate traffic 
at a few aiiports, inevitably those already most constrained by capacity shortages and 
largely located in relatively close proximity within the dominant axis. The most 
important are London Heathrow, Frankfurt, Amsterdam and Paris CDG. The US hub- 
and-spoke model, with its dominant carrier and dedicated terminals and gates, cannot 
be replicated fully in Europe, largely because of existing restrictions on airport capacity. 
Heathrow, for example, has insufficient airport capacity for B A to mount a proper hub- 
and-spoke system, the carrier depending instead on what might be termed random - or 
continuous - hubbing, in which a high degree of connectivity is achieved through its 
sheer volume of flights across the airport. KLM's operation at Schiphol is the most 
fully developed example of a European hub-and-spoke system, having four major and 
two lesser waves per day; the hub serves 120 European and 1 10 long-haul destinations. 
As 609r of the airline's business comes from transfers across this hub. its desire to 
expand European market share and the capacity control policy proposed by the Dutch 
government were obviously in conflict. Although the major European carriers are 
being forced to develop .secondaiy hubs - BA at Gatwick, Air France at Lyon Satolas, 
KLM at Milan Malpensa and Rome (through its Alitalia alliance) - because of 
congestion at their primary bases, they still cannot afford to dilute feed for high yielding 
intercontinental routes - their most profitable services - which depend on maximizing 
the incidence of potential transfers across that core hub. In order to achieve this goal, 
all the major European carriers have established networks of feeder routes increasingly 
operated b\' groups of affiliate regional airlines, which have lower cost structures but 
rai-ely operate aircraft larger than 120 seats (Graham. 1 997b i. 

Feeder routes usually link a secondary city - which may well be m a different 
countr\ - to the intercontinental hub. Such services are escalating in number, partly 
beciiii^c the widespread introduction of regional jets has created a tar more flexible 
produci. .Although hub-feed routes are the most \aluable ser\iccs ttuu regional aiiports 



can provide to consumers in connectivity terms, they also increase movements, 
particularly by smaller aircraft, and inevitably exacerbate airport capacity problems. 
Although regional aircraft generally require only short runways, almost all major EU 
airports lack such dedicated facilities, forcing inefficient use of main runways by small 
aircraft, which also increases ATC problems with aircraft separation. Thus capacity 
constrained airport operators will seek to rationalize demand for their scarce resource by 
adopting pricing staictures that militate against smaller aircraft. For example, regional 
aircraft have largely been forced out of Heathrow. One way to obviate the difficulty of 
hub access is the concept of the aiiport system in which the feeder - or reliever - airport, 
used for regional traffic, might be linked to the hub by dedicated train. One such 
example is Dusseldorf Express (Moenchengladbach), which is being developed as a 
reliever field for slot-constricted Dusseldorf Rhein-Ruhr. Regional carriers may also 
opt for (or be forced to use) secondary airports. KLM uk (formerly Air UK), for 
example, has exploited spare capacity at London's Stansted and City airports, while, 
more generally, capacity-rich secondary aiiports - especially those in downtown 
locations and accessible only to small jets or turbo-props - can provide regionals with 
competitive niche markets for point-to-point traffic. 

The process of hub concentration is being accompanied by an apparently 
contradictory trend towards dispersal as more secondary cities develop international 
routes. The liberalization of transatlantic bilateral agreements has produced a 
'fragmentation' of that market, in which the proliferation of gateways in North America 
and Europe means that many more city-pair markets are served direct by smaller twin- 
jets. This long-haul fragmentation is replicated at theregional scale by the rapid 
expansion of hub-bypass routes, increasingly serviced by regional jets. Although this 
dispersal may have beneficial effects for congestion at individual airports, the increased 
ATMs generated by the additional services compound the negative effects of air 
transport on global air quality. 

2. Frequency as an airline strategic tool 

As in the United States, it is apparent that the hub-and-spoke system evolving in 
Europe, contradicts the argument that the projected growth of demand for air transport 
can, at least in part, be accommodated by the use of larger aircraft. These have better 
seat/mile costs and do offer a means of enhancing capacity at given aiiports without 
increasing dep|artures; they do, however, create more noise. But only one major 
European carrier - BA - is pursuing this strategy, largely reconstructing its Heathrow- 
based fleet around aircraft with a minimum capacity of around 180 .seats. Heathrow 
already has the highest number of passengers per ATM in Europe and the ca.se forT5 is 
that this momentum can be maintained. 

That BA is the exception to the rule is underlined by the statistic that almost 
907f of the aircraft added to the fleet serving intra-European schedules since 1987 are 
less than 1 70 seats: 'Airport congestion has had only a modest influence on airline fleet 
requirements' (Boeing, 1998, p. 28). The implications for aiiport capacity are ■ 
profound. It is readily apparent that the hypothetical use of larger aircraft conflicts with 
the evidence that 'airlines will continue to pursue strategies that accommodate growth 
primarily through additional frequencies' (Boeing, 1997, p. 3). Boeing estimates that 
70% of aircraft deliveries over the next decade will be single-aisle models (mostly less 
than 200 seats), which will account for 71% of the world fleet by 2006, dropping only 
marginally to 69.1% in 2016. Such projections underline the fragility of any argument 
that growth can be partly accommodated in larger aircraft. BA can pursue its strategy 
of increasing aircraft size at Heathrow, only because routes incapable of supporting 
larger aircraft at a sufficient frequency are being diverted to Gatwick. or even 
.Manchester and Birmingham. The Boeing 737s displaced from Heathrow have largely 
gone 10 Gatwick but. then, it too is dependent on increasing aircraft size to meet its 
projecied capacit}' targets. 

The airline fixation wiih iVequeiicy as the priniai-y means ol' accommodating 
growth stems from its role as a. - if not flic - primary form of iion-pncc competition. 
The mix of aiicraft in European aiiline fleets is being dri\en b_\' the need to maximize 
lrec|uency in tiie competitive niarkei-place. market share being iiia\iiiii/.ed bv frequencv 



share, which essentially demands smaller aircraft. Further, competitive market entry 
demands a matching of frequency with that of the incumbent carrier(s). Thus British 
Midland Aii-ways, which in early 1998 began service between Manchester and 
Heathrow in competition with BA, is offering eight daily frequencies but using only 
130 seat Boeing 737-500s. In one sense, this is an indefensible use of very scarce 
capacity in one of the world's most congested airport systems on a city-pair that should 
be served by HSTs. In another, however, it represents a rational business decision that 
reflects the airline's integration as a feeder into the operations of Lufthansa and SAS. 
Nor are long-haul services exempt from the frequency strategy, some airlines on the 
North Atlantic, for example, having down-sized from Boeing 747s to smaller twin-jets 
operating at higher frequency. The importance of frequency is compounded by 
evidence that it is high-yield business-class passengers who are most sensitive to this 
factor. Consequently, most European carriers have linked frequency and status 
products, further reducing the capacity of their aircraft to install separate business-class 
cabins and/or seating for those paying for premium tickets that maximize frequency 
benefits, including the ability to switch flights. The problem is that this behaviour, 
which constitutes rational behaviour for the individual airline, is incompatible with 
wider notions of environmental sustainability. 

Because frequency has evolved as such a key strategic weapon for airlines in 
the competitive market-place, aircraft size has actually declined in certain markets. 
Thus while Air France has radically enhanced frequency on the heavily contested 
domestic trunk routes between Paris, and Marseille, Nice and Toulouse as its response 
to competition from AOM French Airlines and Air Liberie, it is doing so using aircraft 
no larger than 180 seats. As late as the mid-1990s, the most common aircraft on these 
routes were wide-bodied Airbus A300s of Air Inter, carrying over 300 passengers. 
One i-esult is that the average number of passengers per aircraft movement at Paris Orly 
(the principal French domestic aiiport) dropped from 126.8 in 1995 to 108.9 in 1996. 

Although consumers benefit from more frequent services, the negative 
environmental effects of the widespread use of relatively small aircraft (defined as the 
sub-optimal use of scare capacity resources) are compounded by unimpressive load 
factor statistics. Those of European Regions Airline Association members have 
.scarcely changed during the past decade, averaging only 53. 1%. Again, although long- 
haul statistics have improved, at around 64%, the short-haul cross-border passenger 
load factors of AEA members are scarcely higher now than they were in the mid-1980s 
(AEA. 1998). The combination of frequency as a competitive weapon with relatively 
modest load factors means that the 'slot productivity' of many major European airports 
larely exceeds 100 passengers/commercial aircraft movement. Airlines, moreover, are 
forced to try and sell suiplus capacity through special fares and promotions. Such 
tactics, ofcour.se, simply encourage increased mobility and the pressures on scarce 
resources. The operation of Frequent Flyer Programmes (FFPs) has a similar effect in 
that these encourage people to consume mobility which they belie\'e to be free. Perhaps 
as much as 10% of traffic on some airlines is accounted for in this way. leading to 
suggestions that FFPs should be taxed or even banned. 

3. Low-cost airlines 

While the advent of aggressive US-style low cost/'low-no frills' airlines such as Dublin 
and Stansted based Ryanair. Virgin Express at Brus.sels. easyJet at Luton and the BA 
subsidiary. Go. at Stansted. has been hailed as one of the major benefits of European 
liberalization (clearly so for passengers), their aggregate effect has again been to 
increase mobility. Essentially low-fare, point-to-point operators, dependent on low 
costs and high capacity, these airlines may effectively be competing w ith more 
conventional transport modes - classic rail, ferry and long-distance coach - as much as 
incumbent airlines. Their expansion demonstrates that price can create markets, albeit 
largeK located within the regions already most densely seiA'cd hy exisiing carriers. For 
acce.s^ to cheaper and a\ ailablc capacity, the lou-cost operator^ m:\\ ii^e lesser aiiports 
close to major cities, hut their o\erall impact is to contradict principles of sustainability 
111 that the\ contribute to air transport congestion in the EL's doniinaii! axis, while 



13 



encouraging growth in mobility and adding to aggregate air transport emissions and 
noise. 

4. Alliances and code-sharing 

Hub dominance, especially when combined with aiiport congestion, offers the real 
possibility of pre-empting - or at least controlling - competition. Moreover, it is a 
strategy increasingly linked to another - the tactical airline alliance. The EU's hubs are 
also becoming the European centres of the global alliances being orchestrated by the 
world's most powerful airlines (Schiphol, for example, is the European base for the 
KLM-Northwest Airlines grouping). Through acquisition, negotiation and the 
increasingly widespread use of franchising, almost all the most powerful European 
carriers have constructed intra-continental coalitions which, in turn, form part of wider 
global agreements. There are very few small wholly independent airlines in the EU and 
most entrants soon enter into code-sharing, franchising or other agreements with the 
majors. While such strategies are aimed at subordinating the free market to the interests 
of the largest airlines, there may be, perhaps, an inadvertent environmental bonus. By 
their very nature, alliances curb capacity growth and, hypothetically, should allow more 
efficient use of existing resources. This may not be good for competition but adopting 
a different perspective, unconstrained competition in air transport is wasteful of 
investment and resources, including non-renewable hydrocarbons and scarce aiiport 
capacity. It also increases the externalities of air transport, particularly atmospheric 
emissions, noise and terrestrial congestion. It remains to be seen if the rapidly 
escalating incidence of alliances has a beneficial impact on aiiport capacity congestion 
but this is the only current airline tactic in a competitive market-place which offers any 
such potential. 

Conclusions: environmental sustainability, airport capacity and 
European air transport liberalization: towards a resolution? 

The preceding discussion has demonstrated that the different players in the air transport 
industry, making what they regard as the most feasible and judicious decisions 
regarding their own business and consumer interests, are in conflict with each other. 
As many as five sets of essentially irreconcilable tensions define the fault-lines in the 
relationship between environmental sustainability and the air transport industi7 in 
Europe. First, any acknowledgement of the relevance of environmental sustainability 
requires acceptance of the idea that 'infinite mobility is not infinitely good' 
(Bleijenberg. 1995, p. 14), which is the very antithesis of the strategic policies adopted 
by airlines in a deregulated market-place. Secondly, even though a number of airports 
will remain capacity-rich, the aggregate projected growth rates of air transport in 
Europe cannot be accommodated within existing or planned aggregate aiiport capacity, 
particularly when the demand for air transport is geographically concentrated in those 
regions defined by the highest GDP/capita, a spatial pattern unlikely to alter 
significantly. Thirdly, aiiport capacity is essentially being driven by environmental 
criteria, which implies - de facto - if not de jure - constraints on air traffic growth. 
Fourthly, and in marked contrast, the business strategy of the European airline industiy 
in its newly competitive ethos demands network and frequency characteristics, which 
exacerbate the demand for aiiport capacity at a rate even greater than that required by 
aggregate growth in passenger traffic. In these respects, airlines are not behaving with 
due regard for environmental factors, but the corollary is that it would be commercially 
suicidal for any one firm so to do. If, however, the purpose of competition is to 
eradicate competitors in the longer-term, the proces.ses of globalization in the airline 
indListr\- might be viewed as beneficial because their ultimate aim can be inteipreted as 
generating higher profits from capacity control. Finally, the liberalization policy for 
European air transport is arguably at odds with the commitment to 'sustainable 
niohilii\ ' ^upposedly at the heart of European transport policy. 

Ill apparent confirmation ol' the argument that policie> Ui iiiiiigaie the : 
cn\ ironniental impacts of transport are subsumed by counier\'ailing market ■ 
dcvelopiiicnis. the EU lacks an integrated air transport and en\ironmcnial policy. The 
Third Package prioritises competition ai the expense of any oilier goals of social 



14 



solidarity, cohesion or environmental sustainability. Air transport policy is for air 
transport alone, failing to address either the integrated nature of transport itself, or the 
broader concerns and demands of society. This failure to produce an air transport 
policy, which effectively addresses the wider implications of the mode's activities, 
obviously opens up the possibility of other institutional actors becoming involved 
through the application of piecemeal standards, which, ironically, undermine the 
supposedly 'level playing-field' being sought through competition policy. It would be 
foolish of the air transport industry to regard unilateral actions such as the capping of 
Schiphol as isolated cases, for they are more likely to be exemplars that will be 
followed elsewhere, albeit in a haphazard fashion because of the failure to agree post- 
Chapter 3 environmental standards. Society is moving towards an aggregate 
acceptance that infinite mobility cannot be sustained, even if individual people and 
companies are not yet prepared to modify their behaviour accordingly. But the tenor of 
EU environmental policy is explicit; the 'polluter pays' principle - if properly applied - 
implies an internalization of environmental costs through increased direct and indirect 
taxation on transport use and more stringent regulations on noise and emissions. The 
environmental externalities of air transport are thus seen as a market failure to be 
redressed through market mechanisms. The demand for mobility will be suppressed by 
measures aimed at making air transport pay those 'real' costs. Against this, however, 
is the important point that any pricing mechanism for reducing demand is inequitable in 
that business travellers and the wealthy are penalised far less than the members of 
society less able to pay. 

Within that scenario, what actions can the stakeholders in the air transport 
industry take? Some additional infrastructural capacity will be built but it will be 
nowhere near sufficient to meet projected growth. In a rational world, that scarce 
resource would be used more effectively through the deployment of larger aircraft and 
high-capacity one-class cabins (in a very real sense, the European IT industry already 
offers a sustainability model) although - as we have seen - that would flatly contradict 
airline economics. Aiiport charge regimes can penalize small aircraft although this 
actually discriminates against accessibility to peripheral regions and regional airlines. A 
more equitable solution might be to link airport charges to load factor. 

Traffic diversion offers some potential, given that the impact of capacity 
problems varies spatially and the large number of capacity-unrestricted airports in the 
EU, not all of which are sited in remote locations. It is an overstatement to claim that 
'the future is to fly from an airport that no-one wants to fly from to an airport no-one 
wants to fly to' {Air Transport World, May 1996, p. 68). Many travellers are happy to 
use secondary airports for point-to-point journeys, especially if price compensates for 
any inconvenience. Airports such as Manchester and Lyon Satolas can develop 
credible long-haul and connecting scheduled networks, while IT companies are 
prepared to use any regional airports servicing sufficient demand. In addition, general 
aviation activities could be concentrated at reliever fields. However, neither traffic 
diversion or modal shift to HSTs reduce aggregate levels of mobility. 

Ultimately, however, none of these tactics - or even all of them - can solve the 
capacity versus growth equation. Eventually, EU policy-makers must address the 
problem that aggregate mobility in Europe, air transport included, exceeds the 
environmental optimum. It is unlikely that laissez-faire attitudes, which disregard the 
need for an integrated environment-transport policy, will prevail, no matter how much 
the airlines might want this. In many ways, the environmentally-dn\cn strategies of 
aiiport operators are an exemplar of what is to come. The ethos of the times is also 
against traditional command/control environmental policy. Instead, we have the 
Schiphol .scenario in which the aiiport business is allocated overall limits but then 
organizes its own activities within those constraints. Its effectiveness does depend, as 
Bleijenbere ( 1995) argues, on airlines renouncing their apparent preferred objective that 
[here sliould be no en\ ironmental policy at all. Ultiinaiely. any resolution of the 
IlKlni^c^l ^en^ions between en\ ii'Oiimenuil siistauiabiiit)'. airpon capacit\' and European 
air transport liberalization, depends on the de\elopmeni and application of common 
slandardN. Onh' iheii can the behaviour of an individual airline be coiiiniensurate with 



the wider interests and goals of society, without that company being penalized in terms 
of competition. 

References 

AEA - see Association of European Airlines. 
Air Transport World {1996) Tweaking the establishment May, 65-8. 
Association of European Airlines (1996) Yearbook 1996. AEA, Brussels 
Association of European Airlines (1998) Yearbook 1998. AEA, Brussels. 
Aviation Environment Federation (1995) UK airport capacity: AEF response to the 

Transport Select Committee Inquiry, December 1995. AEF, London. 
Aviation Environment Federation (1997) personal communication. 
Banister, D., Capello, R. and Nijkamp, P. eds (1995) European Transport and 

Communication Networks. John Wiley, Chichester. 
Black, W.R. (1996) Sustainable transport: a US perspective. Journal of Transport 

Geography 4, 151-9. 
Black, W.R. (1998) Sustainability of transport. In Modern Transport Geography, eds 

B.S. Hoyle and R.D. Knowies, 2nd. ed., in press. John Wiley, Chichester. 
Bleijenberg, A.N. (1995) A future for air transport. World Transport Policy and 

Practice 1, 12-16. 
Boeing (1997) Boeing: Current Market Outlook 1997. Boeing Commercial Airplane 

Group, Seattle. 
Boeing (1998) Boeing: Current Market Outlook 1998. Boeing Commercial Airplane 

Group, Seattle. 
Button, K. (1994) Privatisation and deregulation: its implications for negative 

transport externalities. Annals of Regional Science 28, 125-38. 
Button, K. and Nijkamp, P. (1997) Social change and sustainable transport. Journal 

of Transport Geography 5, 215-8. 
CAA - see Civil Aviation Authority. 

Cameron, D. (1998) Regulation: help is at hand. Airline Business ianwiry , 36-9. 
CEC - See Commission of the European Communities. 
Civil Aviation Authority (1998) CAP 685 The Single European Aviation Market: The 

First Five Years. CAA, London. 
Collins, A. and Evans, A. (1994) Aircraft noise and residential property values: an 

artificial neural network approach'. Journal of Transport Economics and 

Policy XXWII, 174-97. 
Commission of the European Communities (1994) COM (94) 106 final. Community 

Guidelines for the Development of the Trans- European Transport Network 

EC, Brussels/Luxembourg. 
Commission of the European Communities (1996a) First Report on Social and 

Economic Cohesion, 1996. EC, Brussels/Luxembourg. 
Commission of the European Communities (1996a) Fifth Environmental Action 

Programme: Towards Sustainability. EC, Brussels/Luxembourg. 
Commission of the European Communities (1997a) Towards Sustainability: the 

European Commission's Progress Report and Action Plan on the Fifth 

Progranv)ie of Policy and Action in Relation to the Environment and 

Sustainable Development. EC, Brussels/Luxembourg. 
Commission of the European Communities (1997b) The Single Market Review: 

Subseries 11: vol. 2: Impact on Services: Air Transport. EC, 

Brussels/Luxembourg/KoganPage: Earthscan, London. 
Commission of the European Communities ( 1998) Air Transport and the 

Environment . EC, Brussels/Luxembourg. 
Dennis, N. ( 1 994) Airline hub operations in Europe. Journal of Transport Geographx 
2. 219-33. 
Fergusson. .\!.. Haigh, N. and Hewett. J. I 1994) EU Transport Policy and llie 

Enviro)uiumt: An Overview. Institute for European Environmental Policy, 

London. 
Fli;.^hr International ( 1997) Eurowings adapts services in the face of rail competition 

3-9 December, 12. 



Flight International {1998) Overnight parcel carriers face noisy opposition 15-21 

April, 11. 
Graham, B. (1995) Geography and Air Transport. John Wiley, Chichester. 
Graham, B. (1997a) Air transport liberalization in the European Union: an 

assessment. Regional Studies 31, 807-12. 
Graham, B. (1997b) Regional airline services in the liberalized European Union 

Single Aviation Market. Journal of Air Transport Management 3, 227-38. 
Graham, B. (1998) Liberalization, regional economic development and the geography 

of demand for air transport in the European Union. Journal of Transport 

Geography 6, 87-104. 
Greene, D.L. and Wegener, M. (1997) Sustainable transport. Journal of Transport 

Geography 5, 177-90. 
Hitchens, D. (1997) Environmental policy and the implications for competitiveness in 

the regions of the EU. Regional Studies 31, 813-9. 
Maddison, D. (ed.) (1996) Blueprint 5: The True Costs of Road Transport. 

Earthscan, London. 
Nijkamp, P. and van Geenhuizen, M. (1997) European transport: challenges and 

opportunities for future research and policies. Journal ofTran.sport Geography 
5,4-11. 
Pearce, D. (ed.) ( 1993) Blueprint 3: Measuring Sustainable Development. Earthscan, 

London. 
Pennington, G., Topham, N. and Ward, R. (1990) Aircraft noise and residential 

property values adjacent to Manchester International Airport. Journal of 

Transport Economics and Policy XXIV, 49-59. 
Simon, H-R. (1998), Developing an effective capacity management strategy by 

accurately identifying future requirements. Paper delivered to Aiiport 

Capacity Mangagement Conference, London, 26-7 February. 
Stanners, D. and Bourdeau, P. (eds) (1995) Europe's Environment: The Dobris 

Assessment. Office for Official Publications of the European Communities, 

Luxembourg/European Environment Agency, Copenhagen. 
Thompson, L (1995) High-speed transport hubs and Eurocity status: the case of Lyon. 

Journal of Transport Geography 3, 29-38. 



17 



Table 1 Declared hourly runway capacities for summer busy periods. Source- CAA 
(199S, p. 46). 



Airport 




Capacities 








1993 


1995 


1998 


Median 
1998 


Single Runway 








34.5 


Gatwick 


36-45 


40-47 


42-48 




Manchester 


41 


42 


45-47 




Geneva 


30 


30 


35 




DLisseldorf 


30 


30 


34 




Milan Linate 


24 


22 


32 




Athen.s 


30 


30 


30 




Converging Runways 








52 


Stockholm Arlanda 


63 


66 


70 




Zurich 


60 


60 


66 




Vienna 


30 


45 


54 




Madrid 


35 


35-50 


50 




Barcelona 


28 


30 


47 




Hamburg 


40 


42 


45 




Parailel Runways 








76 


Heathrow 


77-79 


77-81 


75-84 




Paris CDG 


76 


76 


76-84 




Copenhagen 


74 


76 


81 




Munich 


68 


70 


80 




Frankfurt 


68 


70 


76 




Paris Oriy 


70 


70 


70 




Brussels 


53 


60 


64 




Rome Fiumicino 


50 


56 


63 




Milan Malpensa 


30 


30 


26 





IS 



Table 2 Maximum hourly movements at some principal EU airports.' Source- Rolls- 
Royce Market Outlook, 1997. 



Airport 1996 2015 %age increase 



Amsterdam^ 


86 


140 


Athens^ 


32 


80 


Barcelona 


30 


55 


Brussels 


60 


80 


Copenhagen 


69 


80 


DCisseldorP 


36 


65 


Frankfurt 


70 


100 


London Gatwick 


43 


48 


London Heathrow 


82 


85 


Madrid 


43 


50 


Munich 


70 


110 


Paris CDG 


76 


120 


Paris Orly 


70 


80 


Rome Fiumicino 


56 


70 


Stockholm Arlanda 


66 


100 


Zurich 


60 


100 



Median 47.5 



63 

150 
50 
33 
16 

81 
43 
12 
4 
16 
57 
58 
14 
25 
52 
67 



' Projections assume absence of environmentally-driven limits on movements. 

- Theoretical projections; both have current caps on ATMs for environmental reasons. 

^ Assumes completion of new airport at Spata. 




Figure 1 Principal EU airports (plus Norway and Switzerland),. 1996, by 
million passengers handled. Source: Airports Council 
International, Geneva. 



.■ r ■■" ■ ' ~r 




:\i , L.orwa".' an^ 
• r e 3 e d a b 1 ii f u !: u r 



S'.;i czerland) with 
- . Source : Sinon 







t/O C/3 




(U (U 








o o 












o O 


O) 


a. 0-, 


•4-1 

CQ 


■d W) 


•^-t 


o c 


C/D 


c S M 


u 


Qi 


rt i2 rt 


Xi 


i3 cx ^ 


£ 


1 g o- 


Qi 


^ 
^ 


o o ;£ 


Z Z Z, 


\^ 


• • • 





""^^ 


C/3 










c 










o 

• 1— 1 




>, 


c 


c 






o >-, 


o 


o 






'^ o 






^ 




o ;r: 


C/5 


■4-J 

(D 

CX 

6 
o 


tSJJ 

1) 

c 




ortP 
alpo 


s 


O 






O 


o 


rt 




5 ^ 


u 




N 




r2 S 




o 


^^ 




Ph c 




a- 

c 


rt 
n 




mmon ' 
enviro 


o 


re 




^ 


u 


L~ 


o 


W 


r9 -^ 


mJ 


< 


en 


C/) 


u w 


u:i 


• 


m 


• 


• • 





C/5 


CO 




c 
o 






•k^ 


s- 




re 


re 


















oo 




ti) 


1— 




f,i 


*— 1 
















a 


;/; 








o 


C 


r- 


O 


X. 


LU 


• 


• 



tn 
a 



cn 



o 

a. 

B 

o 



u 
o 

Ou 






c 
E 

CJ 

H 

a 

E 

•H 
O 



o 

a 



cn 

(U 

« 
O 

m 

tn 
3 
O 
1-1 
Ul 
B 
> 




> u 










flcS 










QJ _ 










g o 










0) •M 










w 52 










C .S2 


CO 


oo 








^ 

^ 


C 


c 
o 


CO 

1-1 


« a 


■■C 

■ • 


• 




4J 


>:> 6X3 


• 


• 


U .S 










C3 C 










a c 










« a 










^-E 











M 



c 
(1) 
E 
o 



Civil Aviation Development in the Taiwan Area 



By 



Yu-Chun Chang^ and Dr. George Williams^ 



Air Transport Group, 
College of Aeronautics, 
Cranfield University, 
Bedfordshire MK43 OAL, UK 



Fax: +44 (0) 1234 752207 

Tel: +44 (0) 1234 750111 ext. 2242 

E-mail: y.chang. 1998@cranfield.ac.uk 



I .Yu-Chun Chang is a doctoral student at the Air Transport Group, College of 

Aeronautics, Cranfield University. 
2. Dr. George WilUams is a senior lecturer at the Air Transport Group, College 

of Aeronautics, Cranfield University. 



Civil Aviation Development in the Taiwan Area 

By 
Yu-Chun Chang and George Williams 

Air Transport Group College of Aeronautics Cranfield University 

Bedfordshire MK43 OAL, UK 

Fax: +44 (0) 1234 752207 Tel: +44 (0) 1234 7501 1 1 ext. 2242 



Abstract 

Over the past two decades, demand for air transport in Taiwan has grown rapidly, partly in 
response to the economic development of the country, but also as a result of the worsening 
quality of the highway and rail transport systems. Since the deregulation of domestic airline 
services in 1987, the number of routes operated has increased from 20 to 41 and the national 
fleet from 75 to 1 86 aircraft. Over the period, domestic scheduled airline traffic has increased 
at an average armual rate of 18.4%, whilst international scheduled airline traffic has grown at 
10.3%. In 1997, the number of domestic air passengers reached 18.7 million and international 
air passengers 17.1 million. These large increases in demand have resulted in a great impact on 
the use and operation of airport facilities. 

The area of Taiwan is slightly smaller than that of the Netherlands, being some 36,000 sq. km 
(14,000 sq. miles). It is 394 km (245 miles) long and 144 km (89 miles) at its broadest point. 
Seventeen airports serve civil aviation, ten of which are located on the mainland and seven on 
off-shore islands. How is it that a country of this limited geographical area can generate over 
35 million air passengers and 1.2 million tons of air cargo annually? 

This paper reviews and analyses the existing airport facilities, provides a comprehensive 
transport demand forecast and examines the progress and recent development of Taiwan 's 
airline industry. It is based on extensive research carried out by the authors and a team from 
Institute of Transportation, Ministry of Transportation and Communication, Taiwan between 
1995 and 1997. This included desk research, a passenger survey and interviews with the 
airport authorities. It has been updated by the authors to take account of more recent 
developments, especially the Open Skies" agreement with the US and the various airline 
alliances that have occurred involving intemational and domestic carriers. 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



Civil Aviation Development in the Taiwan Area 

1. Introduction 

1.1 Area 

The area of Taiwan is slightly smaller than that of the Netherlands, being some 36,000 sq. km 
(14,000 sq. miles). It is 394 km (245 miles) long and 144 km (89 miles) at its broadest point. 
The eastern side of the island is dominated by a north-south mountain range rising to 13,100 
feet. The western side of the island is a fertile plain through which the country 's only 
motorway links the capital Taipei in the north to Kaohsiung in the south. Surrounding Taiwan, 
there are many small islands including the Penghu Islands, the Kinmen Islands, the Matsu 
Islands, Orchid Island and Green Island. The need for fast links between major cities in 
Taiwan and between these small islands and Taiwan creates opportunities for the air transport 
industry. 

1.2 Population 

The population of Taiwan was a little over 21 million in 1997. It has increased at an average 
annual growth rate of 1.2% since 1983, when it totalled 18.7 million. After 1995 however 
population growth declined to less than 1.0% per annum, reflecting the fact that Taiwan is 
reaching its saturation level in terms of people (see Figure 1-1). 

1.3 GDP 

Taiwan's economic structure has changed considerably since the mid-1980s. High-tech 
products have constituted a sharply increasing percentage of exports. Over the same period, 
the Gross Domestic Product (GDP) increased sharply from US$52.4 billion in 1983 to 
US$283.3 billion in 1997, with an average annual growth rate of 12.81%. Economic growth 
between 1986-1989 was particularly high, averaging more than 20%, with a peak of 35% 
achieved in 1987 (see Figure 1-2). 



2. Air Transport in Taiwan 

2.1 Airports 

There are seventeen airports that serve civil aviation in the Taiwan Area, ten of which are 
located on the mainland and seven on off-shore islands (see Figure 2-1). 

Of the seventeen airports, two are international airports and fifteen domestic. Only CKS 
International, Kaohsiung International and four off-shore island airports (Green Island, Orchid, 
Wangan and Chimeiyu) are exclusively devoted to civil aviation, with the remaining eleven 
domestic airports shared with the military (see Table 2.1). 

2.2 Airlines 

The first privately owned air carrier, Foshing Airiines, was established in 195 1 . FarEastem Air 
Transport and China Airlines followed in 1957 and 1959 respectively. Taiwan A irlines, 
Yu-Chun Chang & George Williams, ATRG Conference, June 1999 2 



Formosa Airlines and Great China Airlines were set up in 1966. By the Mid-1980s, however, 
there were only four air carriers actually operating in Taiwan. China Airlines and FarEastem 
Air Transport operated on domestic trunk routes, while Formosa Airlines and Taiwan Airlines 
focused on off-shore islands routes. Foshing Airlines by then had concentrated on the 
profitable air catering business and also worked as general sales agent for foreign airlines. 

After the proclamation of deregulation of Taiwan's air transport industry in 1987, more and 
more new companies entered the market. Makung Airlines was set up in October 1988. It was 
later renamed UNI air when 5 1% of its shares were sold to EVA Airways. After reorganization 
of their Board members, Great China Airlines and Foshing Airlines resumed their domestic air 
transport services in 1988. Aiming at international operations, Foshing Airlines changed its 
identity and was renamed TransAsia Airways. China Asia Airlines was established in 1989 
and renamed U-Land Airlines when the U-Land Construction Group took over the airline in 
1994. EVA Airways was established in 1989 and entered into the domestic market in October 
1994. Mandarin Airlines, a subsidiary of China Airlines, was formed to operate international 
routes in 1991. Mandarin Airlines is the only airline that does not operate domestic route. A 
summary of Taiwan airlines 'fleets in 1995 is provided in Table 2.2. 

2.3 Domestic Airline Market Supply 

Table 2.3 shows how the supply has changed over the past eleven years. 

i). Number of Airlines: domestic airlines have increased from 4 in 1987 to 9 in 1997. 
ii). Destinations Served: between 1987 and 1997, these have increased from 13 to 16. 
iii). Service Routes: between 1987 and 1997, these have increased from 20 to 41. 
iv). Frequencies Operated: the total number of flights has increased from 76,580 in 1987 to 

286,170 in 1997. 
v). Seats Provided: these have increased from 4.2 million in 1987 to 28 million in 1997. 

2.4 Domestic Air Transport Deregulation Policy 

The Civil Aviation Industry Administration Rules announced in 1949 gave the Civil 
Aeronautics Administration (CAA) of Taiwan the authority to: 

i). Control entry into the industry, 

ii). Control entry into new or existing routes, 

iii). Control exit by requiring approval before cessation of service to a point or on a route, 

iv). Regulate fares, 

v). Control mergers and intercarrier agreements, 

vi). Investigate deceptive trade practices and unfair methods of competition. 

In October 1987, the Ministr>' of Transportation and Communication (MOTC) proclaimed the 
dpen sky ' policy for the domestic air transport industr>-. The policy mainly focused on 
providing a looser control on entry into the industry and of entry into new or existing routes. 
The fare control rule was modified in 1989 to allow airlines to have more freedom to discount 
rates. 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



In 1990, the MOTC prevented further entry into the industry, because there were too many 
domestic airlines operating in the market. 



3. Demand for Air Transport 

3.1 Competition witli Surface Transport 

The highway is the most popular transport mode in Taiwan due to the close proximity of the 
cities. In 1980, 2,060 million passenger journeys were undertaken by road, accounting for 
93.5% of total passenger traffic, while air transport carried only three million passengers, 
equivalent to 0.14% of total traffic. However, the highway system has become saturated due to 
the rapidly increasing use of private cars. The situation is at its worst when there is any public 
holiday. People need to spend more than twice the amount of time on journeys during the peak 
season. It is very crowded on the trains at this time as well, owing to capacity limitations of 
the existing rail network. Any further development of the air transport industry would thus 
have the obvious effect of easing the congestion on surface transport. As a consequence, the 
air system increased its market share to 1.23% of total traffic in 1996, while the highway 
market share declined to 87.53%. The changes in market share of domestic transport modes 
are shown in Table 3.1. 

3.2 Demand for Domestic Passenger Air Transport 

The domestic air transport industry has experienced a growing trend since 1971 (see Figure 3- 
1), when the number of passengers carried by airlines was just over one million. Between 1972 
and 1978 the industry grew rapidly with an average annual growth rate of 25.35%. In 1978 the 
number of passengers carried by airlines was nearly four million, about four times greater than 
the figure of seven years earlier. However, after 1979 the demand for air transport declined 
continuously for four years at an average rate of 8.36%. This was mainly caused by 
improvements in the ground transport system, including the completions of the first motorway, 
northeast railroad and the electric powered railway system. The improvements in the ground 
transport systems led passengers away from the airiine industry. In 1980 the passengers carried 
by airlines decreased by 19%, while those carried by the railway system increased by more 
than 8%). The substitution effect of demand between air and ground transport in Taiwan was 
thus apparent. 

Between 1985 and 1986, the passenger demand again declined. The main reason for this was 
that China Airlines had put most of its resources into expanding its international routes. For 
the domestic market, the airline kept only three B737 to operate six domestic routes. The 
capacity of these routes thus remained almost unchanged for many years. The other reason 
was that FarEastem Air Transport had problems due to confusion among the main shareholders 
which affected the operation of the airiine. Most of the B737s operated by the company were 
introduced about 20 years ago. As a result of the problems experienced in the 1980s, there was 
no updating of the fleet. After 1987, the domestic air transport industry of Taiwan returned to 
an era of rapid growth following deregulation. The demand for air transport increased sharply 
from 1987 to 1997, with an average growth rate of 1 8.35%. 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



3.3 Demand for and Supply of Domestic Passenger Air Transport 

One of the major reasons for deregulation was to develop the capacity of domestic air transport 
so as to ease the congested ground transport in Taiwan. This aim has been achieved in the 
deregulation era. Under deregulation, airlines may easily enter into new markets and increase 
the capacity of routes. 

The demand for domestic air transport increased 27% in 1987, whilst the capacity of the 
industry increased by only 17%. The gap between supply and demand resulted in a high seat 
load factor, reaching more than 80%) (see Figure 3-2). With such a high occupancy, passengers 
frequently experienced difficulty in booking seats on their desired flights. During the peak 
season, especially Chinese New Year, passengers had to queue for hours to buy tickets. It was 
apparent that the industry 's supply was far below the public 's demand. This was the main 
reason behind the government 's decision to deregulate domestic air transport. 

After the deregulation of domestic air transport, the growth rate of annual available seats 
exceeded the growth rate of annual passenger demand. This was mainly due to the fact that 
airlines started to introduce new larger size aircraft into the market. Total available seats 
numbered 4 million in 1987, but by 1997 had increased to 28 million. The total increase in 
annual available seats was nearly 700% over the ten-year period. The number of available 
seats has boomed since 1991. The average growth rate of annual available seats has been 
nearly 24% over the last six years. This rapid increase in supply resulted in excess capacity 
and created a more competitive market environment in the industry. 

3.4 Competition at the Route Level 

In 1987, more than 95% of routes were operated by one or two carriers. Of the 20 routes 
operated, eight were served by one carrier and eleven were operated by two carriers (see Table 
3.2). Since the new carriers entered into the market the situation has changed. By 1996, the 
routes served by one carrier had declined to 23%) of the total number of routes served, whilst 
the routes served by four or more carriers had increased to 20%). It is clear that the operation of 
the air transport industry before deregulation was either monopolistic or duopolistic. The 
domestic air transport industry after deregulation however is much more competitive. The 
increased competition in the market may be explained by the increasing number of airlines. 
The continued existence of monopoly in 1 997 is mainly caused by the expansion of operations 
to smaller airports. These small airports usually have only short runways, which require 
airlines to operate aircraft with short takeoff and landing capabilities. This generates a certain 
kind of natural barrier to other carriers who do not have such type of aircraft. For example, 
Formosa Airlines is a monopolistic operator on the Taipei-Matsu route, because its Domier 228 
is the only aircraft allowed to operate on such a short runway as that at Matsu. This kind of 
technical barrier cannot exist for long however. When the runway extension program is 
completed, more airlines will enter the market. 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



3.5 Market Share of Carriers 

Prior to deregulation there were only four carriers operating domestic routes. China Airlines 
and FarEastem Air Transport were the two major players operating the trunk routes in 1986, 
carrying nearly 90% of the total passenger traffic. Formosa Airlines and Taiwan Airlines were 
two minor operators who focused on the thin routes and carried the remaining 10% of traffic 
(see Table 3.3). 

FarEastem Air Transport was the most important domestic air carrier, with a market share of 
over 50%. Confusion among its main shareholders adversely influenced the company 's fleet 
update however, leading to a decline in its market share from 57% in 1986 to 29% in 1997. 

As Formosa Airlines and Taiwan Airlines operated mainly on the thin routes linking small 
islands with Taiwan, they avoided head on competition with the big two carriers. Small 
propeller aircraft were widely used by both these airlines to suit the operational requirements of 
the short runways on the islands. TransAsia Airways entered the market in 1988 and grew 
rapidly. Its market share was only 1% in 1988, but by 1995 it had over 27%. Since then its 
share of traffic has declined to around 22%. Great China Airlines and Makung Airlines were 
the other two major airlines to enter the domestic air market in 1989. Each had their market 
share increase from less than 5% in 1989 to more than 10% in 1997. U-Land Airline achieved 
notoriety in Taiwan when in 1994 the U-Land Construction Group took over China Asia 
Airlines, adopting a marketing strategy to fly Taipei-Kaohsiung route with one NT dollar". 
The strategy proved very successful for U-Land Airline as it was able to increase its market 
share from 0.057% in 1994 to 6.1 1% in 1997. 

3.6 Market Shares by Route 

The Taipei-Kaohsiung route is the busiest consistently accounting for more than 30% of the 
domestic market over recent decades (see Table 3.4). Before 1992, the Kaohsiung-Makung 
route was the second busiest route as a result of Makung, the largest off-shore island in the 
Taiwan Area, being poorly served by ferry. After 1992, demand on the Taipei-Tainan route 
increased sharply, raising it from being the fifth to the second busiest route in the domestic 
market. 

The Taipei-Kinmen route is the other one which has grown rapidly. After deregulation of 
services to Kinmen Island in 1990, the demand for air transport increased sharply with the 
route 's overall market share rising from 0.52% in 1987 to 5.45% in 1995. 

3.7 Air Fares 

Air Fares remained under the CAA 's control after deregulation. Airlines had to get CAA 
approval before issuing any new fare. Air Fares have been raised three times to cover 
increased operational costs over the past ten years (see Table 3.5 and 3.6). Beuveen 1990 and 
1993 fares on most routes increased by around 12%, reflecting the effects of inflation, although 
in certain cases there was no change in the rate changed. In the period 1993-1995, fare 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



increases varied between 4% and 1 1%. 

In December 1995, the CAA proclaimed a change to air fare policy to allow airlines to have 
more freedom to adjust their fares freely within a maximum discount rate of 30%. On special 
occasions, such as the inauguration of a new airline, the introduction of a new aircraft, the 
operation of a new route or a company 's anniversary, airlines have the right for special 
promotions within a maximum discount rate of 50% for a two week period. Airlines need to 
report to the CAA for such special promotions 30 days before the promotional date. This 
became the first occasion that the CAA loosened its control on air fares since deregulation. 



4. Supply of Airports 

4.1 International Airport Facilities 

Table 4.1 lists the major facilities of the two international airports in Taiwan. There are two 
parallel runways at CKS International Airport, both of which are more than 3,300 meters in 
length. Currently there are 22 in-contact and 8 remote parking stands available for passenger 
aircraft, with 12 parking stands provided for the cargo terminal. 

The other international airport at Kaohsiung operates with one runway. It has 12 in-contact 
parking stands at its passenger terminal and 4 parking stands at the cargo terminal. 

4.2 Airport Traffic Data 

CKS International Airport is the most important international airport in Taiwan, handling 
107,822 aircraft movements, 14 million international passengers and one million tons of 
international cargo in 1997 (see Table 4.2). 

Taipei Airport is the most important domestic airport in Taiwan, with 187,998 aircraft 
movements, 15 million domestic passengers and 39,596 tons of domestic cargo handled in 
1997. 

Kaohsiung International Airport is the second most important airport for domestic services, 
with more than 9 million passengers and 21,057 tons of cargo handled in 1997. 

4.3 Changes in International Airport Operations 

Between 1990 and 1997 traffic growth at CKS International Airport was moderate, with a 
9.82% annual increase in aircraft movements, a 6.84% annual rise in international passengers 
and a 9.19% annual increase in international cargo (see Table 4.3 to 4.5). 

Traffic at Kaohsiung International Airport has risen sharply since 1990. The number of aircraft 
movements was only 4,000 in 1990, but by 1997 this figure had increased to nearly 27,000. 
International passenger traffic increased nearly four times between 1990 and 1997, with an 
average growth rate of 21.33%. The average growth rate of international cargo was 18.95% 
between the same years. 

Yu-Chun Chang & George Williams, ATRG Conference, June 1999 7 



4.4 Changes in Domestic Airport Operations 

Most of the domestic routes serving Taipei are handled at Sung Shan Airport, located 5 
kilometers north of the city centre. Sung Shan is a hub airport and is the busiest for domestic 
routes, serving more than 15 million passengers in 1996. 

Traffic at Taichung, Chiayi, Tainan, Kaohsiung and Pindung Airports has increased rapidly due 
to the congested ground transport. These five airports are located in the west corridor of 
Taiwan, where most of the population is concentrated. Not surprisingly, these west corridor 
routes have become the most popular in the domestic market. The average annual growth rate 
of passenger traffic at these five airports was more than 20% from 1986 to 1996 (see Table 
4.6). 

Hualien and Taidung are located in eastern Taiwan. As there is no motorway in the east 
corridor, railway and air became the main transport modes for these two cities. The average 
annual growth rate of passenger traffic at these two airports was more than 10% from 1986 to 
1996. 

After the Kinmen route was deregulated, most tourists changed their destinations from Makung 
to Kinmen. As a consequence, traffic at Kinmen Airport has increased sharply from 1993, with 
an average annual growth rate of more than 45% from 1991 to 1996. With such a rapid 
increase in traffic the new passenger terminal built in 1991 is now too small. 

4.5 Air Transport Forecast 

According to the I ATA Air Transport Forecast of 1997 (see Table 4.7), total domestic 
passenger traffic in Taiwan grew by 26% per annum on average between 1990 and 1995. The 
strong growth experienced in the last five years in domestic travel was the result of the 
development of services by several regional carriers following liberalization. The growth in 
international traffic was much more moderate, with an 8.7% annual rate experienced between 
1990 and 1995 (see Table 4.7). 

lATA anticipated that domestic passenger traffic would grow faster than international traffic 
between 1995 and 2000. While domestic passenger traffic is expected to grow by 12.7% per 
annum between 1 995 and 2000, international scheduled passenger traffic to and from Taiwan 
should grow by 10.9% per annum between 1995 and 2000 and 7.1% per annum thereafter. 



5. Recent Air Transport Developments in Taiwan 
5.1 APROCPIan 

The Asia-Pacific Regional Operations Center Plan (APROC) is the key to Taiwan 's economic 
future. Whether Taiwan can respond to change, break through bottlenecks and occupy a 
significant place in the global economy of the 21" centur>' all largely depend on this plan. One 
of the important APROC aims is the establishment of the Air Transport Centre, including 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



express air-cargo transit hub and air-passenger transit hub. 

5.1.1 Express Air-Cargo Transit Hub 

A special area for express cargo operations will be planned and set up in the cargo terminal at 
CKS international airport. International express-cargo operators will be allocated their own 
exclusive operating areas within the airport. They will be permitted to install and operate their 
own high-efficiency equipment. A policy of liberalizing commercial air-cargo operations will 
be carried out. The ground services company at CKS intemational airport will be privatized 
without delay and ground handling operations will be opened up to a second operator. 

The second phase of the cargo-terminal extension project at CKS intemational airport will be 
completed, a special zone for express cargo will be created and operating capacity will be 
expanded. 

Development of the airport as an intemational express cargo transport hub will be accompanied 
by the full integration of storage, carriage, information technology, manufacturing and other 
related activities. 

5.1.2 Air-Passenger Transit Hub 

In the short term, to make Taiwan more attractive to transit passengers: 

i). The first phase of the plan to extend and improve the passenger terminal at CKS 

intemational airport will be carried out. The space for resting and shopping will be 

expanded and the quality of service will be raised, 
ii). The issuance of visas on arrival and the privilege of visa-free entry vsdll be extended and 

custom clearance will be made more rapid and efficient, so as to render it more 

convenient for passengers to stop over in Taiwan. 

In the mid to long term, to build up the physical infrastructure and make every effort to expand 
and develop passenger transit operations: 

i). The second phase of the CKS intemational airport terminal extension and development 
plan will be vigorously pressed ahead with. Commercial areas and rest facilities will be 
greatly increased, 
ii). The airport 's ground transport links will be improved. Long-distance bus services to and 
from the airport will be opened up to a second operator. A rapid transit network will be 
built to connect the airport to Taipei. Air routes will be opened for cormecting flights to 
central, southem and eastern Taiwan. 

To strength management and organization: 

i). In the short-term, corporate-management practices will be introduced. Changes will be 

made to the organization and functions of the Civil Aeronautics Administration to 

strengthen its operational efficiency. 
ii). The second phase of the terminal project will be put under private-sector management, 
iii). The airport 's commercial operations, such as hotel accommodation, shops, restaurants. 

Yu-Chun Chang & George Williams, ATRG Conference, June 1 999 9 



cafeterias, parking lots and the maintenance of the terminal facilities, will be assigned 
to private-sector management. 

5.2 Open Skies Agreement with the U.S. 

In March 1997, Taiwan signed an Open Skies 'agreement with the U.S., Taiwan being the 16th 
country to sign such an agreement with U.S. 

After the signing of the Open Skies ' agreement with the U.S., the two main international 
airlines, EVA Airways and China Airlines, entered into alliances with Continental Airlines and 
American Airlines respectively. EVA Airways began its codeshare agreement with Continental 
Airlines in March 1998, with Continental Airlines codesharing on EVA Airways ' flights from 
Los Angeles, San Francisco, Seattle, Newark and Honolulu to Taipei. In turn, EVA Airways 
will codeshare on Continental 's flights throughout the U.S. By linking flight schedules, the 
two carriers will greatly reduce flight connection times between the U.S. and Asia. The 
agreement also enables Continental and EVA to offer reciprocal frequent flier programs, shared 
airport lounges and through check-in to final destinations. 

China Airlines began its codeshare agreement with American Airlines in December 1997, with 
China Airlines codesharing on American Airlines' services from Los Angeles and San 
Francisco to Dallas, Chicago, Miami, New York and Washington D.C. In turn, American 
Airiines will codeshare on China Airlines ' flights to Taiwan. The agreement also enables 
China Airiines and American Airlines to offer reciprocal frequent flier programs, shared airport 
lounges and through check-in to final destinations. 

5.3 Domestic Airline Alliances 

Airline Alliances have become popular in the domestic market of Taiwan in recent years. They 
first appeared in the 1980s when China Airiines acquired 19% of FarEastem Air Transport's 
shares. With the benefit of alliances, airlines provide passengers with more flexible choices by 
enabling them to take alliance partner airiines ' flights using the same tickets. 

Recent alliance activities have occurred since the purchase of 24% of the shares of Great China 
Airiines and 43% of the shares of Makung Airlines by EVA Airways in 1995. By purchasing 
shares, EVA Airways has built up close alliance relationships with Great China Airiines and 
Makung Airlines, which have benefited from receiving EVA's support on crew training, 
maintenance, service, ticketing image. EVA benefited by rapidly expanding its domestic 
network, acquiring a number of feeders for its international routes, increasing its domestic 
market share and acquiring precious slots at some congested airports. All airiines benefited 
from reduced operating costs through sharing facilities and by ordering the same type of 
aircraft. Great China, Makung and EVA together have ordered the MD90, getting a much 
better price in the process. EVA Airways went on to expand its alliance activity to include 
Taiwan Airlines, purchasing 29.74% of its shares in 1996. 

China Airlines followed the trend by forming an alliance with TransAsia Airways on the 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 10 



Taipei-Kaohsiung route in 1995. China Airlines further expanded its alliance activity by 
purchasing 33% of the shares of Formosa Airlines in June 1996. 

With the alliance benefit, the most competitive Taipei-Kaohsiung route has become closer to an 
oligopoly market of three major groups instead of the original competitive market formed by 
seven operators. China Airlines, TransAsia Airways and Formosa Airlines form one alliance, 
which took 41% market share in 1995. EVA, UNI, Great China and Taiwan Airlines form 
another alliance, taking 21% market share in 1995. FarEastem Air Transport, the dominant 
carrier in the market, continues to operate independently and took 38% market share in 1995. 
U-Land Airlines was the only small airline which did not join any alliance and took less than 
0.2% market share in 1995. 



6. The Current and Future Environment in the Asia-Pacific Region 

6.1 Rank of Major Asian Airports 

6.1.1 Air Passengers 

In 1996, CKS International Airport ranked 8th of the selected major ten airports in Asia and 
50th in ACI world airport statistics in terms of passenger traffic. If only international 
passengers traffic is included, however, CKS International Airport was ranked 6th of the 
selected ten major airports in Asia (see Table 6.1). 

6.1.2 Air Cargo 

In 1996, CKS International Airport's air cargo traffic ranked 5th of the selected major ten 
airports in Asia and 18th in ACI airport ranking. Developing CKS International Airport into an 
Air Cargo Transit Hub is therefore easier to accomplish than the plan to develop it into an Air 
Passenger Transit Hub (see Table 6.2). 

6.2 Top Ten Asia-Pacific Domestic City-Pairs 

The number of scheduled seats available on the Taipei-Kaohsiung route in 1996 was 8,525,804, 
making it the busiest domestic city-pair in the Asia-Pacific region (see Figure 6-1). 

6.3 Air Transport Forecast for the Asia-Pacific Region 

Taiwan's total air passenger average annual growth rate was 15.4% between 1985-1995, 
ranking it 3rd of the major Asia-Pacific countries. According to lATA 's air transport forecast 
of 1997, Taiwan's domestic passengers will reach 104.1 million in 2010 and its international 
passengers 52.8 million. Total air passenger average annual growth rate is estimated to be 
8.8% over the period 1995-2010 (see Table 6.3). 

6.4 Composition of Asia-Pacific Region Traffic to and from Taiwan 

Figure 6-2 shows the past and future composition of Asia-Pacific traffic to and from Taiwan. It 
can be seen clearly that Northeast Asia will remain the most important region for traffic to and 
from Taiwan, although its share will decline to 49% in 2010, compared to 56% in 1995 and 

Yu-Chun Chang & George Williams, ATRG Conference, June 1999 11 



80% in 1985. 

Traffic between Taiwan and the Americas achieved particularly high rates of growth in the 
recent past, with a 21% annual average rate achieved between 1985 and 1995. However, Asia- 
Pacific was by far the most important world region for international traffic to and from Taiwan, 
accounting for 87% of total international passengers in 1995. 

6.5 International Air Traffic to and from Taiwan 

In 1985, Japan was the most important country for international traffic to and from Taiwan, 
followed by Hong Kong. But in 1995, Hong Kong became the most important market for 
intemational traffic to and from Taiwan. This occurred because there is no direct service 
between Taiwan and China (see Figure 6-3). 

According to the lATA air traffic forecast of 1997, the assumed introduction of direct 
scheduled services between Taiwan and China in 1998 will make the Taiwan-China route area 
become the second most important route for Taiwan in 2010 after Hong Kong. 

6.6 Top Ten Asia-Pacific Countries in terms of Domestic Passengers 

In 1985, Japan was by far the most important Asia-Pacific domestic travel market, while 
Taiwan was ranked fifth. By 1995, Japan remained the most important Asia-Pacific domestic 
travel market, but domestic passenger traffic in China and Taiwan had grown rapidly and 
ranked second and third respectively. According to the I ATA air transport forecast of 1997, by 
2010, China will be the major Asia-Pacific domestic travel market, followed by Japan and 
Taiwan (see Figure 6-4). 



7. Conclusion 

The domestic air transport industry in Taiwan hais experienced a growing trend since 1971 due 
to the changed economic structure and the worsening quality of surface transport. In 1987, the 
Ministry of Transportation and Communication (MOTC) proclaimed the dpen sky 'policy for 
the domestic air transport industry. The policy provided a looser control on entry into the 
industry and of entry into new or existing routes. As a result, the demand for domestic air 
transport grew rapidly with an average annual growth rate of 18.35% between 1987 and 1997. 

Over the past ten years, domestic airlines have increased from four in 1987 to nine in 1997, 
destinations served have increased from 13 to 16, routes operated have risen from 20 to 41, the 
total number of flights provided has increased from 76,580 to 286,170 and the total number of 
seats offered has increased from 4.2 million to 28 million. 

In 1987, 95% of the routes operated were served by one or two carriers, but by 1996, the routes 
served by only one carrier had declined to 23% of the total number of routes served, whilst the 
routes served by four or more carriers had increased to 20%. It is clear that the operation of the 
air transport industr>- before deregulation was either monopolistic or duopolistic. Since 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 12 



deregulation however, the market is much more competitive. The increased competition in the 
market may be explained by the larger number of airlines. 

CKS International Airport is the most important international airport in Taiwan, handling 
107,822 aircraft movements, 14 million international passengers and one million tons of 
international cargo in 1997. Its air passenger and air cargo traffic ranked 8th and 5th 
respectively of the selected major ten airports in Asia and 50th and 18th respectively in ACI 
world airport rankings in 1996. 

Taipei Sung Shan Airport is a hub airport and is the busiest for domestic routes, serving more 
than 15 million passengers in 1996. The Taipei-Kaohsiung route is the busiest domestic city- 
pair in the Asia-Pacific region. 

The Asia-Pacific Regional Operations Center Plan (APROC) is the key to Taiwan 's economic 
future. One of the important APROC aims is the establishment of the Air Transport Centre, 
which will include an express air-cargo transit hub and an air-passenger transit hub. A special 
area for express cargo operations will be planned and set up in the cargo terminal at CKS 
international airport. 

In March 1997, Taiwan signed an Open Skies 'agreement with the U.S., Taiwan being the 16th 
country to sign such an agreement with the U.S. After the signing of the Open Skies ' 
agreement, the two main international airlines, EVA Airways and China Airlines, entered into 
alliances with Continental Airlines and American Airlines respectively. 

Airline Alliances have become popular in the domestic market of Taiwan in recent years since 
the purchase of 24% of the shares of Great China Airlines and 43% of the shares of Makung 
Airlines by EVA Airways in 1995. China Airlines followed the trend by purchasing 33% of 
the shares of Formosa Airlines in June 1996. With the benefit of alliances, airlines provide 
passengers with more flexible choices by allowing them to take alliance partner airlines 'flights 
using the same tickets. 

Taiwan's total air passenger average annual growth rate was 15.4% between 1985-1995, 
ranking it 3rd of the major Asia-Pacific countries. According to the lATA Air Transport 
Forecast of 1997, Taiwan's domestic passengers will reach 104.1 million in 2010 and its 
international passengers 52.8 million. Total air passenger average annual growth rate is 
estimated to be 8.8% over the period 1995-2010. 

Northeast Asia will remain the most important region for traffic to and from Taiwan, although 
its share will decline to 49% in 2010, against 56% in 1995 and 80%o in 1985. In 1985, Japan 
was the most important country for international traffic to and from Taiwan, followed by Hong 
Kong. But in 1995, Hong Kong had become the most important country for international 
traffic to and from Taiwan. According to lATA 's assumed introduction of direct scheduled 
sen'ices between Taiwan and China in 1998, the Taiwan-China market will become the second 
most important route for Taiwan in 2010 after Hong Kong. 

Yu-Chun Chang & George Williams, ATRG Conference, June 1999 13 



References 

APEC Transport Working Group (1996), "Congestion Points Study Phase II, Volume 2: Air 

Transport", February. 
China Airlines (1997), 'China Airlines and American Airlines Enter into Strategic Alliances", 

China Airlines News, November. 
Feng, Cheng-Ming (1998), "Recent Development of Taiwan's Regulatory Changes in 

International Air Transport", Journal of Air Transport Management, 4, pp. 165-1 67. 
Hufbauer, Gary Clyde and Findlay, Christopher (1996), "Flying High- Liberalizing Civil 

Aviation in the Asia Pacific", Institute for International Economics, Washington, DC, 

November. 
lATA (1997), "Asia-Pacific Air Transport Forecast 1980-2010", January. 
Lee, Gwo-Chian (1996), "The Impact on Airlines of Deregulation in Taiwan", MSc Thesis, 

Cranfield University, September. 
Taiwan, Business Times, (1998), "Taiwan Signed Open Skies Agreement with American", 

January 22. 

Taiwan, Civil Aeronautics Administration, MOTC (1998), Civil Aviation Statistics Annual 
Report. 

Taiwan, Civil Aeronautics Administration, MOTC (1999), 'Air Transport Policy White Book", 

January. 
Taiwan, Council for Economic Planning and Development (1995), "A Plan for Building 

Taiwan into an Asia-Pacific Regional Operations Center", Executive Yuan. 
Taiwan, Institution of Transportation, MOTC (1993), "Development of an Air Transportation 

Hub in the Taiwan Area", June. 
Taiwan, Institution of Transportation, MOTC (1995), "A Study in the Economic Regulation of 

Domestic Transport in Taiwan Area", October. 
Taiwan, Institution of Transportation, MOTC (1997), "A Study of Civil Aviation Development 

in Taiwan Area" July. 
Taiwan, Institution of Transportation, MOTC (1998), "A Comparative Study of Major 

International Airports in the Asia Pacific Region", July. 
Taiwan, Institution of Transportation, MOTC (1998), The Third Taiwan Area Integrated 

Transportation Systems Planning", July. 
Taiwan, Institution of Transportation, MOTC (1998), Transport Data Analysis Annual Report. 
Taiwan, Institution of Transportation, MOTC (1998), Transport Economic Data Statistics 

Aimual Report. 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 ' 14 



Table 2.1 Categorization of Taiwan Airports Operating International and Domestic Services 



Function 


Characteristic 


Airports 


Number 


International 
Airports 


Civil only 


CKS International, Kaohsiung 
International 


2 


Domestic 
Airport 


Military and Civil 
Aviation 


Taipei, Hsinchu, Taichung, Chiayi, Tainan, 
Pingtung, Hualien, Taidung, Makung, 
Kinmen, Matzu 


11 


Civil only 


Green Island, Orchid, Wangan, Chimeiyu 


4 



Table 2.2 The Fleets of Taiwan fe Scheduled Airlines (1995) 



Airline 



Aircraft 



Average Seating 
Capacity 



Fleet 
Size 



Total Fleet 
Size 



Operating Routes 



China Airlines 



MD-11 
A 300 
A 320 
B737 
B747 



275 


4 


254 


12 


150 


2 


117 


3 


360 


19 



40 International and 

Domestic 



Madarin Airways 


B 747-409 


411 


1 


1 


International 


EVA Airways 


B 767-300ER 


238 


9 


25 


International and 




B 767-200 


212 


4 




Domestic 




B 747-400 


386 


6 








MD-11 


275 


6 






Trans Asia 


ATR-42 


50 


■ 3 


23 


International and 


Airways 


ATR-72 


74 


12 




Domestic 




A 320-231 


162 


6 








A321-131 


194 


2 






Far East Air 


B737 


120 


4 


15 


International and 


Transport 


B757 


210 


2 




Domestic 




MD-82 


154 


8 








MD-83 


165 


1 






Great China 


DHC-8-102 


39 


2 


14 


Domestic 


Airlines 


DHC-8-311 


56 


12 






UNI Airways 


HS-748 


54 


2 


8 


Domestic and 




BAe-146 


112 


5 




International 




B757 


210 


1 




Charter 



Formosa 


SAAB-340 




36 


9 


Airlines 


Fokker-50 




56 


4 




Fokker-100 




109 


1 




Domier-228 




19 


7 




BN-Islander 




16 


2 




UH-12EHeli 


copter 


2/1 


3 



26 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



Domestic 



Taiwan 

Airlines 




BN-2A 
Domier-228 


9 
19 


4 
1 


5 


Domestic 


U-Land 
Airlines 




MD-82 
SHORTS 360 


154 
89 


2 

1 


3 


Domestic and 

International 

Charter 


Source: CAA, 


MOTC 


, 1996. 











15 



Table 2.3 Supply Changes in the Domestic Airline Market 



Year 


No. of 


Destinations 


Routes 


Frequencies 


Seats Provided 




Airlines 


Served 


Operated 


Operated 




1987 


4 


13 


20 


76,580 


4,199,591 


1988 


5 


13 


21 


80,266 


4,812,844 


1989 


7 


13 


22 


107,492 


5,699,800 


1990 


8 


13 


23 


110,163 


5,947,741 


1991 


8 


13 


24 


132,782 


7,862,187 


1992 


8 


13 


25 


148,051 


9,731,694 


1993 


8 


14 


27 


176,815 


13,275,809 


1994 


9 


14 


29 


215,663 


16,468,880 


1995 


9 


16 


33 


237,458 


22,022,033 


1996 


9 


16 


35 


284,749 


27,027,765 


1997 


9 


16 


41 


286,170 


27,980,000 



Source: CAA, MOTC, 1998. 



Table 3.1 Changes in Market Share of Domestic Transport Modes (%) 



Year Highway Rail Air 



1980 93.50 6.36 0.14 

1981 93.70 6.16 0.13 

1982 93.73 6.14 0.13 

1983 93.73 6.14 0.13 

1984 93.98 5.88 0.14 

1985 94.03 5.83 0.13 

1986 93.79 ' 6.08 0.13 

1987 93.43 6.40 0.16 

1988 93.19 6.62 0.19 

1989 92.89 6.86 0.25 

1990 92.20 7.53 0.27 

1991 91.99 7.70 0.32 

1992 91.12 8.47 0.41 

1993 90.56 8.93 0.51 

1994 89.61 9.70 0.68 

1995 88.56 10.51 0.93 
1996 87^3 11.23 1.23 



Source: Transport Data Analysis, JOT, MOTC, 1987-1997. 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 16 



Table 3.2 Competition in City-Pair Routes 



Year 


One Carrier 


Two Carriers 


Three Carriers 


Four or More 
Carriers 


Total 




















Routes 




Routes 


% 


Routes 


% 


Routes 


% 


Routes 


% 


1987 


8 


40.00 


11 


55.00 





0.00 


1 


5.00 


20 


1988 


8 


38.10 


9 


42.86 


1 


4.76 


3 


14.29 


21 


1989 


8 


36.36 


9 


40.91 


2 


9.09 


3 


13.64 


22 


1990 


8 


34.78 


9 


39.13 


3 


13.04 


3 


13.04 


23 


1991 


9 


37.50 


9 


37.50 


3 


12.50 


3 


12.50 


24 


1992 


8 


32.00 


12 


48.00 


1 


4.00 


4 


16.00 


25 


1993 


10 


37.04 


11 


40.74 


3 


11.11 


3 


11.11 


27 


1994 


9 


31.03 


11 


37.93 


6 


20.69 


3 


10.34 


29 


1995 


11 


33.33 


12 


36.36 


6 


18.18 


4 


12.12 


33 


1996 


8 


22.86 


17 


48.57 


3 


8.57 


7 


20.00 


35 


1997 


13 


31.71 


16 


39.02 


5 


12.20 


7 


17.07 


41 



Source: CAA, MOTC, 1998. 



Table 3.3 Changes in Market Shares (%) of Scheduled Airlines 



Year 


China 
Airlines 


FarEastern 

Air 
Transport 


Trans Asia 
Airways 


Great 

China 

Airlines 


Formosa 
Airlines 


Taiwan 
Airlines 


Makung 

Airlines 

(UNI) 


China Asia 
Airlines 
(U-Land) 


EVA 
Airways 


1986 


31.60 


57.37 


- 


- 


6.53 


4.50 


- 


- 


- 


1987 


36.09 


52.41 


- 


- 


7.41 


4.09 


- 


- 


- 


1988 


37.07 


51.53 


1.00 


- 


■7.64 


2.76 


- 


- 


- 


1989 


29.06 


43.22 


6.66 


3.28 


10.52 


2.98 


4.28 


- 


- 


1990 


20.13 


44.52 


11.01 


6.06 


10.06 


2.91 


5.18 


0.13 


- 


1991 


16.32 


37.74 


16.00 


8.16 


8.04 


2.18 


11.58 


0.06 


- 


1992 


11.04 


43.08 


18.69 


8.74 


6.95 


1.48 


9.82 


0.18 


- 


1993 


8.51 


43.40 


21.97 


9.30 


6.76 


0.91 


9.15 


0.01 


- 


1994 


6.48 


40.17 


26.26 


9.60 


7.98 


0.63 


8.53 


0.06 


0.29 


1995 


4.59 


37.96 


27.45 


10.10 


8.63 


0.68 


7.47 


0.15 


2.96 


1996 


4.30 


30.39 


25.08 


9.87 


10.75 


0.85 


9.95 


4.72 


4.08 


1997 


4.85 


29.04 


21.64 


10.89 


11.51 


1.22 


11.48 


6.11 


3.26 



Source: CAA, MOTC, 1998. 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



17 



Table 3.4 Changes in Market Share (%) by Route 



Route 



1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 



Taipei-Kaohsiung 


32.01 


32.76 


33.54 


34.63 37.07 


39.61 39.07 


37.52 


37.18 


40.94 


Taipei-Tainan 


6.24 


5.99 


6.3 


7.38 


9.05 


10.19 11,4 


12.02 


11.49 


11.84 


Taipei-Hualien 


13.54 


13.22 


11.84 


11.34 


9.27 


8.18 7.17 


7.05 


6.86 


6.39 


Kaohsiung-Makung 


14.15 


15.11 


13.64 


12.77 11.15 


9.29 7.99 


6.50 


5.80 


4.93 


Taipei-Taidung 


4.14 


4.26 


5.05 


5.60 


5.20 


5.30 5.23 


5.30 


5.45 


4.89 


Taipei-Kinmen 


0.52 


2.09 


2.32 


2.64 


2.66 


3.25 5.12 


5.71 


4.95 


3.85 


Taipei-Makung 


10.54 


10.46 


10.72 


10.36 


9.88 


7.86 6.67 


5.23 


4.50 


3.68 


Kaohsiung-Hualien 


3.53 


3.53 


3.18 


3.10 


2.66 


2.39 1.84 


1.73 


1.85 


1.86 


Taichung-Makung 


2.19 


2.29 


2.24 


2.24 


2.35 


2.45 2.06 


1.79 


1.60 


1.41 


Tainan-Makung 


2.16 


1.92 


2.46 


2.32 


2.31 


2.01 1.74 


1.52 


1.30 


1.00 


Table 3.5 


' Changes 


i in Air Fares (NT Dollars) of Inland Routes 


I 






Route 


1 


1990 


1993 


1996 




Increase % from 


Increase % from 














1990-1993 


1993-1995 




Taipei-Kaohsiung 




1209 


1323 


1409 




9.43 




6.50 


■^^^ 


Taipei-Tainan 




1110 


1239 


1325 




11.62 




6.94 




Taipei-Chiayi 




1050 


1181 


1272 




12.48 




7.71 




Taipei-Taichung 




840 


945 


1023 




12.50 




825 




Taipei-Hualien 




919 


1028 


1111 




11.86 




8.07 




Kaohsiung-Hualien 




1335 


1428 


1511 




6.97 




5.81 




Taichung-Hualien 




1820 


1900 


1975 




4.40 




3.95 




Taipei-Taidung 




1226 


1323 


1407 




7.91 




6.35 




Kaohsiung-Taidung 




1004 


1130 


1214 




12.55 




7.43 




Taichung-Taidung 


== 


1960 


1960 


2036 





0.00 




3.88 





Table 3.6 Changes in Air Fares (NT Dollars) of Off-shore Island Routes 



Route 



1990 



1993 



1996 



Increase % from 
1990-1993 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



Increase % from 
1993-1995 



Taipei-Makung 


1044 


1164 


1252 


11.49 


7.56 


Kaohsiung-Makung 


736 


821 


909 


11.55 


10.72 


Tainan-Makung 


699 


779 


867 


11.44 


11.30 


Taichung-Makung 


1044 


1051 


1091 


0.70 


3.81 


Chiayi-Makung 


721 


804 


897 


11.51 


11.57 


Taipei-Kinmen 


1544 


1544 


1629 


0.00 


5.51 


Kaohsiung-Kinmen 


1234 


1376 


1461 


11.51 


6.18 


Taidung-Orchid 


990 


1104 


1154 


11.52 


4.53 


Taidung-Green Island 


495 


552 


602 


11.52 


9.06 


Kaohsiung-Orchid 


1395 


1555 


1607 


11.47 


3.34 


Kaohsiung-Chimeiyu 


1170 


1305 


135S 


11.54 


4.06 


Chimeiyu-Makung 


580 


647 


701 


11.55 


8.35 


Kaohsiung-Wangan 


1195 


1332 


13S5 


11.46 


3.98 



18 



Table 4.1 Major Facilities of Taiwan 6 International Airports 



Facilities 


CKS International 
Airport 


Kaohsiung 
International Airport 


Runway 


Main Runway 


Number 


2 


1 


Length x Width (m) 


3660 x60 
3350 x60 


3150 x60 


Aided Runway 


Number 


1 


1 


Length x Width (m) 


2752 x45 


3050 x45 


Passenger 
Terminal 


Total Floor Area (m^) 


163,900 


68,200 


Stands 


In-contact 


8747x10 
Wide Body x 12 


B747 X 4 
Wide Body x 8 


Remote 


B747 X 8 


- 


Parking Slots 


Bus 


55 


30 


Car 


2,098 


1,140 


Taxi 


150 


238 


Cargo 
Terminal 


Total Floor Area( m^) 


94,180 


50,900 


Stands 


B747 X 4 
Wide Body x8 


B747 x 4 


Parking Slots 


Truck 


115 


36 


Car 


520 


215 



Source: lOT.MOTC, 1998. 



Table 4.2 Airport Traffic Data (1997) 



Airport 


Aircraft Movements 


Passengers 


Cargo (tons) 




Domestic 


International 


Domestic 


International 


Domestic 


International 


CKS 


1,096 


107,822 


21,560 


14,163,294 


420 


1,099,745 


Kaohsiung 


114,711 


27,908 


9,223,316 


2,905,388 


21,057 


104,184 


Taipei 


187,998 


- 


15,394,038 


- 


39,596 


- 


Hualien 


28,300 


- 


1,855,722 


- 


3,521 


- 


Taidung 


39,870 


- 


1,398,643 


- 


3,691 


- 


Makung 


51,044 


- 


2,124,330 


- 


13,692 


- 


Taichung 


50,402 




1,878,247 


- 


4,837 


- 


Tainan 


26,790 




2,496,419 


- 


4,397 


- 


Chiayi 


22,771 




1,043,695 


- 


2,134 


- 


Chimeiyu 


5,583 




43,861 


- 


424 


- 


Wangan 


1,217 




8,714 


- 


113 


- 


Orchid 


5,433 




66,719 


- 


326 


- 


Green Island 


13,684 




162,394 


- 


392 


- 


Kinmen 


19,320 


- 


1,397,638 


- 


7,709 


- 


Matzu 


7,141 


- 


103,008 


- 


821 


- 


Pingdung 


4,755 


- 


181,255 


- 


521 


- 


Total 


580,115 


135,730 


37.399,559 


17,068,682 


103,651 


1,203,929 


Source: CAA, 


MOTC, 1998. 













Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



19 



Table 4.3 Changes in Aircraft Movements 



Year 


CKS International Airport 


Kaohsiung International Airport 


Total 




Movements 


Percentage 


Movements 


Percentage 


Movements 


1990 


56,537 


93% 


4,154 


7% 


60,691 


1991 


62,080 


91% 


6,101 


9% 


68,181 


1992 


68,982 


89% 


8,560 


11% 


77,542 


1993 


74,451 


85% 


13,479 


15% 


87,930 


1994 


83,409 


83% 


16,974 


17% 


100,383 


1995 


92,195 


82% 


19,599 


18% 


111,794 


1996 


101,371 


82% 


22,560 


18% 


123,931 


1997 


108,918 


80% 


26,812 


20% 


135,730 


Annual Ave. 
Growth Rate 


9.82% 




30.53% 




12.19% 



Table 4.4 Changes in International Passenger Traffic 



Year 


CKS International Airport 


Kaohsiung International Airport 


Total 




Passengers 


Percentage 


Passengers 


Percentage 


Passengers 


1990 


8,929,218 


92% 


750,701 


8% 


9,679,919 


1991 


9,356,836 


90% 


1,007,462 


10% 


10,364,298 


1992 


10,827,878 


89% 


1,289,395 


11% 


12,117,273 


1993 


11,153,612 


87% 


1,733,041 


13% 


12,886,653 


1994 


11,618,574 


85% 


2,054,325 


15% 


13,672,899 


1995 


12,585,798 


84% 


2,401,781 


16% 


14,987,579 


1996 


13,585,851 


84% 


2,570,947 


16% 


16,156,798 


1997 


14,184,854 


83% 


2,905,388 


17% 


17,090,242 


Annual Ave. 
Growth Rate 


6.84% 




21.33% 




8.46% 



Table 4.5 Changes in International Air Cargo Traffic 



Year 


CKS International Airport 


Kaohsiung International Airport 


Total 




Tons 


Percentage 


Tons 


Percentage 


Tonnage 


1990 


594,642.8 


95% 


30,788 


5% 


625,430.5 


1991 


634,389.9 


95% 


35,737 


5% 


670,127.0 


1992 


723,490.1 


95% 


39,818 


5% 


763,307.8 


1993 


742,729.4 


94% 


50,534 


6% 


793,263.5 


1994 


746,781.6 


92% 


62,672 


8% 


809,453.4 


1995 


941,411.7 


92% 


78,385 


8% 


1,019,796.6 


1996 


986,640.4 


92% 


91,402 


8% 


1,078,042.3 


1997 


1,100,165.3 


91% 


103,763 


9% 


1,203,928.7 


Annual Ave. 
Growih Rate 


9.19% 




18.95% 




9.81% 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



20 



Table 4.6 Domestic Air Passenger Traffic by Airport (Thousands) 



Airport 


1986 


1987 


1988 


1989 


1990 


1991 


1992 


1993 


1994 


1995 


1996 


Ave 

Growth 

Rate 


Taipei 


1,839 


2,347 


2,794 


3,326 


3,430 


4,191 


5,929 


7,438 


9,609 


11,802 


15,204 


23.6% 


Taichung 


55 


79 


98 


148 


137 


179 


350 


525 


800 


1,181 


1,596 


40.0% 


Chiayi 


67 


71 


70 


67 


62 


93 


164 


288 


546 


755 


1,002 


31.6% 


Tainan 


260 


294 


313 


403 


450 


644 


884 


1,198 


1,565 


1,830 


2,356 


24.7% 


Kaoshiung 


1,385 


1,779 


2,057 


2,388 


2,445 


3,039 


3,903 


4,726 


5,675 


6,989 


8,055 


19.3% 


Pingtung 


- 


- 


- 


- 


- 


- 


- 


- 


16 


224 


246 


292.1% 


Hualien 


528 


598 


672 


709 


697 


703 


805 


878 


1,095 


1,343 


1,595 


11.7% 


Taidung 


290 


399 


364 


456 


472 


492 


613 


846 


885 


1,102 


1,173 


15.0% 


Orchid 


- 


- 


- 


- 


- 


48 


93 


76 


79 


78 


63 


5.6% 


Green 


- 


- 


- 


- 


- 


48 


108 


125 


148 


162 


131 


22.2% 


Island 


























Makung 


877 


1,114 


1,246 


1,413 


1,366 


1,534 


1,670 


1,833 


1,781 


1,971 


2,061 


8.9% 


Chimeiyu 


- 


- 


- 


- 


- 


24 


48 


56 


50 


23 


45 


13.4% 


Wanan 


- 


- 


- 


- 


- 


10 


18 


18 


14 


17 


3 


-22.4% 


Matzu 


- 


- 


- 


- 


- 


- 


- 


- 


62 


89 


90 


20.5% 


Kinmen 


- 


- 


- 


- 


- 


196 


290 


626 


914 


1,204 


1,279 


45.5% 


Total 


5,300 


6,680 


7,610 


8,910 


9,060 


11,200 


14,880 


18,640 


23,240 


28,770 


34,896 


20.7% 



Source: Transport Data Analysis, lOT, MOTC, 1987-1997. 



Table 4.7 Air Traffic Forecast for the Taiwan Area 



Domestic Flights 



International Flights 



Total 



Year 



Passengers Growth Rate Passengers Growth Rate Passengers Growth Rate 
(millions) (millions) (millions) 



1985 


5.85 


- 


4.80 


- 


10.65 


- 


1990 


9.04 


9.1% 


10.43 


16.8% 


19.47 


12.8% 


1995 


28.74 


26.0% 


15.82 


8.7% 


44.56 


18.0% 


2000 


52.20 


12.7% 


26.53 


10.9% 


78.74 


12.1% 


2010 


104.13 


7.1% 


52.76 


7.1% 


156.89 


7.1% 



Source: "Asia-Pacific Air Transport Forecast 1980-2010", lATA, January 1997. 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



21 



Table 6. 1 Passenger Traffic at Major Asian Airports (1 996) 



Rank of 


ACI Airport 










Asian 


Ranking 


Airport 


Total 


Internationa! 


Domestic 


Airports 












1 


9 


Seoul Kimpo 


34,706.158 


14,705,015 


19,736.711 


2 


17 


Kai Tak 


30.212,327 


29,542,500 


. 


3 


24 


Narita 


25.408,779 


22,665,870 


794,729 


4 


26 


Bangkok 


24,992,738 


16,380,434 


6,530,554 


5 


27 


Chang-I 


24.514,248 


23,129,802 


- 


6 


42 


Kansai 


18,849,164 


10,095,871 


8,222.544 


7 


44 


Beijing 


16,383,225 


3,909,970 


12.473,255 


8 


50 


CKS 


15,613,624 


13,585,851 


. 


9 


- 


Shanghai 


12,344,826 


• 


* 


10 


71 


Manila 


11,938,454 


7,297,108 


4,641,346 



Source: Airport International July/ August 1997. 

Note; Total passengers includes arriving, departbg and transit. 

* The original data for Shanghai Airport was collected with total passengers and cannot be 
separated into international and domestic passengers. 



Table 6.2 Air Cargo at Major Asian Airports (1996) 



Rank, of Asian 
Airports 


ACI Airport 
Ranking 


Airport 


Cargo (tons) 




1 


5 


Narita 


1,625,840 




2 


6 


Kai Tak 


1,590,772 




3 


9 


Seoul Kimpo 


1,361,510 




4 


12 


Chang-1 


1,190,457 




5 


18 


CKS 


796.155 




6 


22 


Bangkok 


787,539 




7 


- 


Kansai 


592,557 




8 


- 


Manila 


393,344 




9 


- 


Beijing 


390,098 




10 


- 


Shanghai 


304,977 





Note: - means the airport was not shown in the statistics of ACI. 

Source: "Asia-Pacific Air Transport Forecast 1980-2010". lATA, January 1997. 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



22 



Table 6.3 Asia-Pacific Region Major Countries Air Transport Forecast (iVIIIIions) 



Country 


1985-1995 




2010 




1995-2010 




Average Annual 


Domestic 


International 


Total 


Average Annual 




Rates of Growth 


Passengers 


Passengers 




Rates of Growth 


China 


22.9% 


229.1 


62.3 


291.4 


10.9% 


Japan 


7.0% 


134.0 


91.6 


225.6 


4.4% 


Taiwan 


15.4% 


104.1 


52.8 


156.9 


8.8% 


Korea 


16.7% 


54.0 


54.9 


108.9 


7.8% 


Australia 


6.9% 


61.4 


32.5 


93.9 


6.5% 


Thailand 


12.6% 


22.3 


49.1 


71.4 


8.1% 


Hong Kong 


11.3% 


0.0 


70.8 


70.8 


6.5% 


India 


4.7% 


38.3 


22.1 


60.4 


7.2% 


Singapore 


9.6% 


0.0 


56.1 


56.1 


6.6% 


Indonesia 


12.0% 


22.6 


28.8 


51.4 


8.2% 


Malaysia 


10.1% 


13.0 


33.5 


46.5 


6.7% 


Philippines 


6.0% 


17.4 


18.3 


35.7 


7.9% 



Source: "Asia-Pacific Air Transport Forecast 1980-2010", LATA, January 1997. 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



23 



Figure 1-1 Changes in Taiwan 6 Population from 1983 to 1997 




1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 

Year 



Figure 1-2 Changes in Taiwan fe GDP from 1983 to 1997 



GDP (USSbillions) 
300 



250 



I GDP .^-Growth Rate 



200 



Growth Rate 
40% 




1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 

Year 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



24 



Figure 2-1 The Location of Airports and Airline Routes in the Taiwan Area 



(S> 


Tslpfei^irport 

_..••■■"' IntemationaJW. . \ ^ 
/Airport /jfjl\ ' v. ^ 


f 4 

i Matzu Airport 


/@ ///// 1 
1/ // 1 1 ' 


1 


"Sc^ "'••::f:; .•^""^ 

Kinm en Airport / •.. Ta'ipiungAf 


^/ / II ! Huali^ Airport; 

¥ // ' '■' ^ \ 


i s4^ f M/ 

: Makung Airt5ort. I @ / / 

: -^ = ''\Jchiay\ AM)(/n 
• JyWangaif. Airport a. /lit 

j ^ W }'■■// 

: Chiftietyu Airpirt-. J \ / \\ 1 


r / 1 ' '' 

1 \.' ! J 
/ -■ ■' ( 


' 1 1 


■■•••... V>., \ Taihan\Airpdy/^y ^^. 

'■•■••..>•>.••. '\ I ir*© 

■ .■;•■■■.•;•.. ■••. ) -if Pingtijng-Airporl 

■•••"•'.'.■.•.•.•■.l^,.-..r..:. 

Kaohsiu^a'"--... 

Intematiorhii,^^ ""• 

Airport \. 


lunaPsLirpofft / / 

'•%1 #••■ / 

j^ Cr^e^ Island / 

■■■■/'" ■•.._ Airport / 


Legend 




; 


Orchid 


.^rport 






W 




- • - ■ East- West Routes 








Off-Shore Islands Routes 

















Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



25 



Figure 3. 1 Changes in Domestic Air Transport Passenger Demand 



Passengers (millions) ^ Passengers (million) -^ Growth Rate 








Growth Rate 


20 




Rno/. 


18 


t 


■ 


- 50% 


16 


l\ 11 


1 


40% 


14 


1 I /\ 


-1 












. 30% 


12 


1 1 / \ A 


*. 

A--H 


^ 




10 

e 


/ \ / \ / VA / 










20% 
- 10% 


6 


• • y \ / 










. 0% 


4 


- 1 iV/ . 1 1 1 1 










-10% 


2 




....■iiiiriiiiiiiilll 










-20% 




v-ojn^i/itor^aooiov-rjnvintoro-nmo^ 
O)cncna)o>mo)cn<na>o>tno)f»o>cncnoic»0)cn 


rg r» -V irt to r 
<n o> ot o> o) c 
o> o> CD oi ai c 


1 




Year 





Figure 3.2 Changes in Demand for and Supply of Domestic Passenger Air Transport 



Numbers (millions) [^-Annual Seats -♦-Annual Passengers H*-Seat Factor 
30 




1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 

Year 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



26 



Figure 6-1 Top Ten Asia-Pacific Domestic City-Pairs in 1996 



Taipei-Kaohsiung 

Seoul-Pusan 
Sapporo-Haneda 

Seoul-Cheju 

Haneda-Fukuoka 

Melbouene-Sydney 

Okinawa-Haneda 

Cheju-Pusan 
Kuala Lumper-Penang 

Cebu-Manila 




2 3 4 5 6 7 

Annual Scheduled Seats (both directions) (millions) 



Source: "Asia-Pacific Air Transport Forecast 1980-2010", lATA, January 1997. 



Figure 6-2 Past and Future Composition of Asia-Pacific Region Traffic to and from Taiwan 



Percentage 




Year 



Asia-Pacific Region Air Traffic to and from Taiwan 



Source; "Asia-Pacitlc Air Transpon Forecast I9S0-2010", lATA, January' 1997. 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



27 



Figure 6-3 International Air Traffic to and from Taiwan 



Annual Passengers 
(millions) 







2000 



Year 



^*- yl.o'" 



\o' 



.40<^' 



International Air Traffic to and from Taiwan 



Source: "Asia-Pacific Air Transport Forecast 1980-2010", lATA, January 1997. 



Figure 6-4 Top Ten Asia-Pacific Countries in terms of Domestic Passengers in 1985, 1995 and 2010 
Source: "Asia-Pacific Air Transport Forecast 1980-2010", lATA, January 1997. 



Passengers (millions) 




Countries 



Year 



Yu-Chun Chang & George Williams, ATRG Conference, June 1999 



28 



S"' AIRTRANSPORT RESEARCH GROUP (ATRG) CONFERENCE 1999, IN HONG KONG 

Airport Choice & Competition - a Strategic Approach 



Benedikt N. Mandel 

MKmetric Gesellschaft fur Systemplanung mbH, 

Durlacher AUee 49, 

76131 Karlsruhe, FRG 

email: mandel@mkni.de 



Abstract 



Keywords 



After discussing some background information of the aviation sector, a 
strategic approach for the airport management is shown, which can be 
embedded in the decision making process. Two modelling issues are 
highlighted. The one concerns the non-linearity of consumer behaviour which 
touches a major principle in modelling including the airport choice and the 
other issue deals with competition in the aviation sector focusing on the 
possibility of measuring it so that a quantification is possible. 

The applied system approach ensures that effects based on the synergy of air 
networks, the competition among air alternatives and between air and the land 
based modes as well as the co-operation of modes are taken into account m the 
simulation process. Therefor a consistent simulation instrument is available to 
forecast effects of supply changes on travel demand. 

The selected strategic analyses sho%vn in the last section are based on 
elasticities subject to the strategic simulation instrument VIA to forecast 
effects of supply changes on travel demand. The non-linear approach uses 
point as well as cross elasticities of the demand side with respect to supply 
characteristics. 

Air transport, strategic simulation, startegic supply changes, airport choice, 
travel demand, discrete choice, consumer behaviour, non-lineanty, asymmetry, 
threshold, Box-Cox Logit, multimodality, competition of modes, intermo- 
dality co-operation of modes, intramodality, competition of airports or air 
services, elasticity of demand, trip purposes, market shares, catchment-area, 
location of airports, passenger charges, aircraft fees, airport pricing strategy, 
access / egress choice, system approach, management strategies. 



Airport Choice & Competition - a Strategic Approach 2 
TABLE OF CONTENTS 

1. BACKGROUND 5 

2. SYSTEM APPROACH 8 



3. NON-LINEARITY AND THE CONSUMER'S CHOICE 12 

3 . 1 Properties of the linear standard model 12 

3.2 The Box-Cox device 14 

3.3 Visual and economic significance 14 

3.4 Asymmetry 15 

3 .5 Other considerations 1 5 



4. COMPETITION CONCERNING AIR TRANSPORT 18 



5. SELECTED STRATEGIC ANALYSES 23 

5.1. Market situation 23 

5.1.1. Catchment area 24 

5.1 .2. Regional differences 28 

5.2. Interdependencies 33 

5.2. 1 . Location of an airport 33 

5.2.2. Accessibility of an airport 36 
5.3 Scenarios 40 

5.3.1 Route INAUGURATION 41 

5.3.2 Secondary hub 44 

5.3.3 Consumer's ELASTICITY 47 

5 .3 .4 Local pricing strategy 50 



Airport Choice & Competition - a Strategic Approach 



TABLE OF FIGURES 

Fig. 1. System Approach 8 

Fig. 2. Steps of air transport forecast 9 

Fig. 3. Linking the models in the forecasting system 10 

Fig. 4. Linear Logit versus Box-Cox Logit 13 

Fig. 5. Market shares of Frankfurt 1991: intercontinental destinations 24 

Fig. 6. Passengers at Frankfurt by origin for intercontinental destinations 1991 25 

Fig. 7. Market shares at four different Airports; forecast 2004 27 

Fig. 8. Airport choice, destination: North America, trip purpose: business 28 

Fig. 9. Airport choice, destination: Rome (Italy), trip purpose: vacation 29 

Fig. 10. Airport choice from the county of Hanover to different destinations 1991 

30 

Fig. 11. Airport choice from Bielefeld to Hong Kong, full fare business class, 1994 

31 

Fig. 12. Comparing locations for an international airport for Berlin 33 

Fig. 13. Passengers Berlin 2010 depending on airport's location 34 

Fig.l4. Modal split between Berlin and Munich 2010 depending on the location 
of an airport 35 

Fig. 15. Change of passengers mileage depending on an airports location 36 

Fig. 16. Market share of rail Frankfurt Airport 1991 38 

Fig. 1 7. Passengers to Frankfurt Airport by rail 1991 39 

Fig. 18. Passengers (in thousand per year) by train to/from Leipzig Airport, 2015 

40 

Fig.l9. Market shares of Frankfurt Airport: destination North America 42 

Fig. 20. Market shares changes of Frankfurt Airport by regions if non-stop flights 
to North America are offered at the Berlin airport system 43 

Fig. 21. Market share changes at Frankfurt Airport: destination Asia and Africa 

45 

Fig. 22. Market share changes at Frankfurt Airport: destination Asia and Africa 

46 



Airport Choice & Competition - a Strategic Approach 4 

Fig. 23. Share of travel demand depending on cost changes 4g 

Fig. 24. Travel demand depending on cost changes ^g 

Fig. 25. Market share losses of Hamburg Airport 1991: all destinations 51 

Fig. 26. Passengers shifts j2 

Fig. 27. Passengers at Hamburg Airport by trip purpose 1991 53 
Fig. 28. Change of aircraft movements Hamburg Airport 1991 (percentages) 54 

Fig. 29. Aviation revenues depending on passenger charges 55 



TABLES 

Table 1. Airport choice from Bielefeld to Hong Kong, full fare business class, 
1994 



32 



Table 2 Access/egress choice road j 7 

Table 3 Access/egress choice rail j7 



Airport Choice &. Competition - a Strategic Approach 



1. BACKGROUND 

The ongoing world wide process of liberalisation and deregulation is 
accompanied by increasing privatisation in the air industry. Obviously this 
new situation has fundamental impacts and needs consideration in strategic 
planning. Hence, the necessity is given to know more about existing 
competition and - in prospective context - possibilities to assess it so that 
different market situations can be evaluated and incorporated into 
management strategies. 

While airlines faced the rules of competition already for some time the 
airports were mostly excluded due to a number of reasons.' Airlines react to 
the new competitive situation with large productivity optimisation programs 
to cut down costs and the lease of aircrafts to enlarge the short-term financial 
flexibility. In parallel, they defend their markets by using marketing 
strategies like frequent flyer programs, lounge membership schemes or in 
future by the planned project 'virtual airline' as well as by establishing 
international alliances (code sharing, cross-share holding, franchising). 

The concentration of the supply side by the alliances reduces the 
competition so that e.g. in Europe almost ^vo thirds of the existing 0-D's are 
monopoly services, one quarter of the routes is served by two carriers (which 
often belong to the same alliance) and only on the remaining routes there are 
three and more competitors. In the last case - which covers a third of the 
total passenger volume transported - the consumers benefited firom a 
significant drop in price in the past. 

Further competition could be expected due to further liberalisation, new 
entrants offering low-cost services, established 'national' airlines extending 
their businesses to other European areas, increasing capacity constraints and 
improved high speed services. 

The situation at most airports is different because they are still owned by 
the public and with some exceptions (hubs and privatised airports) the 
necessity for competitive behaviour was not given. While the privatisation 
process is going on, subsidies are cut down and productivity has to be 
increased, secondary airports withdraw passengers from the major ones, 
capacity constraints exist at a lot of airports and huge investments are 
requested to solve existing problems. Furthermore airline alliances redirect 
passenger flows to secondary hubs due to the higher prices and / or capacity 
constraints at major hubs as well as based on the optimisation of the air 



' These includes restrictive bilateral air agreements, the lack of deregulation and privatisation, 
the procedures to define aircraft fees and passenger charges, the available capacity 
resources and the missing or small sensitivity of the public to environmental effects, etc. 



Airport Choice & Competition - a Strategic Approach 6 

alliance supply in total. In addition various aiq^orts offer special services on 
the non-aviation side. 

The latter could be interpreted as airports are starting to act like private 
companies in a competitive market. They have to fight for customers, 
increase their attractivity for passengers as well as for airlines and cope with 
competition at the airport itself (e.g. ground handling). Marketing and 
market analysis becomes more important to be able to react fast and 
precisely to market changes and to develop strategies for mid- and long-term 
investments. If the management fails (e.g. with the pricing strategy or the 
infrastructure investments) well established airports of today will be 
downgraded by the consumers and airlines to second or third league airports 
of tomorrow. 

The European air market is facing additional policy interventions in form 
of new deregulations and regulations. There is international pressure to 
follow an open sky policy^ and to cut down subsidies rigorously to allow fair 
competition. Further on politicians are approached by the public to 
internalise external costs due to increasing environmental sensitivity. 

Fees or taxes on aircraft emissions and demand-based aircraft fees as 
well as passenger charges or quotas for air movements and noise are applied 
or taken into consideration. Additional interventions are expected to 
harmonise the market conditions with respect to airport cost structures, local 
landing fees and passenger charges. 

Demand and supply should be the only forces in a free market but the 
existing access to the market conditions needs some additional rules to 
transform e.g. the restrictive grandfather rights on the slot allocation side 
into an open system where slots can be traded. The necessity to install such 
rules is given in the light of the capacity constraints faced by nearly any big 
airport and the slot blocking politics specially of home carriers at hubs, 
which prevent new competitors from entering the market. 

New regulations are welcomed by privatised airports as opportunities to 
maximise their revenues by optimising resource management. Therefore 
peak pricing could be used to cope with capacity constraints - but one has to 
ensure by price caps that airports do not withdraw monopoly rents 
extensively and that new entrants will have a fair chance. So scarce 
resources at congested airports handled at market price will lead to shifts to 
other airports or land-based modes. In parallel, aircraft fee structures can be 



• Non-European carriers should have 'unrestricted' access to the European air market. 
Naturally such harmonisation issues have to be in-line with other actions, for example the 
assignment of a landing point to a specific airport in a bilateral agreement discriminates all 
other airports which are not considered. 



Airport Choice & Competition - a Strategic Approach 7 

used as instruments to meet quotas of noise and aircraft movements.^ 

Airlines and airports have to show great flexibiHty in adjusting their 
supply to the changing demand and regulatory framework in order to survive 
in the evolving market. Therefore the decisions for long- and mid-term 
strategies become more and more difficult and decisive as they might 
comprise costly investments in air-related infrastructure (including high 
speed railway stations at airports), market oriented pricing to handle sparse 
resources or to develop new markets, new services in the non-aviation sector 
or the extension of hub & spoke versus point-to-point air schedules. 

The resulting complexity in decision-making processes in the air industry 
requires enhanced planning instruments to apply appropriate means from the 
administrative side and to adjust supply structures that will enable the 
carriers and airports to stand the increasing competition. The following 
sections will direct towards a strategic planning instrument which helps to 
face the complex problems stated, so that the managers can enrich their 
knowledge by analyses and scenarios to reduce the entrepreneurial risk in 
decision making. Enhanced econometric models which analyse and explain 
possible consumer reactions on adjusted supply figures offer the opportunity 
to study the interdependencies in the market structures, to anticipate future 
changes and to evaluate the resulting effects on both - the microeconomic 
and the macroeconomic level. 

Being the public transport mode with the highest increase in demand 
during the last years an appropriate instrument for analysing air services 
impacts is therefore highly recommended. Unfortunately air traffic is a fairly 
complicated mode to handle - from a modellers point of view. 

The unimodal approaches to assess changing market situations are more 
or less sophisticated extrapolations of the past and it turned out that they are 
poor predictors. Forecast processes that take additionally into account other 
alternatives which are accessible from a consumer's point of view will 
predict in a better way. Hence, multimodal approaches which incorporate 
inter- and intramodal as well as synergetic network effects are per se more 
sufficient to cope with real life complexity. 

Although the approach which will be shown is much richer, within this 
paper it is not possible to deal with all possibilities and all problems. There- 
fore the obvious ones like long term forecasting based on different socio- 
economic, infrastructure and transport policy scenarios will stand aside. Here 
the focus is on the consumer side which at least decide due to their 
sensitivity where the business goes to. Some of the results shown in the final 
section will indicate that the scope of the underlying studies was much 
wider. 



^ Modified aircraft fees could force airlines also to increase their load factors above the 
miracle barrier of 70 per cent. 



Airport Choice & Competition - a Strategic Approach 



2. SYSTEM APPROACH 



A consequent development of this scope leads to a systemic view of 
transportation. It is therefore necessary to embed air transport forecasting in 
a in framework of relevant relationships that include and take into account 
the whole transport market as well as demographic, economic, political, 
spatial and technical components. Figure 1 gives an idea of the considered 
determinants. 



Population 
Economy 



E 
® 



TechnologYi 
InfrastrOdture" 
■ Motorizatioh" 



Urbaiiisatidh'H.^ 
Land use -'r- 




Fig. 1. System Approach 

A modelling process based on these interrelationships explains the 
transport market by multimodal and multisectoral determinants. This 
approach ensures the consistency of the whole model system in every step of 
the simulation process. Considering detailed exogenous impact factors as 
population, economic and political circumstances, technical development 
and spatial structures the models always process balanced figures of all 
endogenous measures. Hence, no transport activity appears or disappears 
unexplained within the system. Changes in the system's state are substitutive 
or complementary and synergetic effects, as well as competition, lead to new 
situations concerning diversion, accessibility or attractiveness. These effects 
can be analysed with respect to modes (e.g. road, rail, sea, air) and/or trip 
purposes (e.g. business, vacation, private). 

In the light of the complex problems stated above it is obvious that the 
airport choice model has to be embedded in some sort of model explaining 
total trip making by all modes and a sort of model explaining the choice of 
mode for a trip. It is convenient to postulate, for the sake of discussion, the 
existence of an aggregate generation-distribution model: this corresponds to 
frequent practice and the points that should be made about an ideal 
specification also hold when disaggregate generation-distribution 
specifications are used. In addition the existence of an disaggregate mode 



Airport Choice & Competition - a Strategic Approach 9 

choice model based on a logistic function, say a logit-model, has to be 
assumed so that the consumer elasticities with respect to the alternative 
modes can be identified. Some additional models are needed to face the 
problems of access/egress choice to the airports and slot choice to explain the 
consumers selection of departure time. Last but not least, assignment 
procedures are required to compute impedances which reflect the 
attractiveness of each alternative based on the infrastructure networks of all 
modes. Figure 2 shows the stages of air transport forecasts and the context of 
the different models. 



E 
© 




Assignment 



i\ccess/egress 



Airport choice 



Mode choice 



Generation/distribution 



Scenario ^ ^ 



definitibfis^ 

-.''iTSSi*' 



Socioeconomic data 



Infrastructun 



mm 



Fig. 2. Steps of air transport forecast 



To encounter the effects from one decision level to the other, say from 
mode choice to generation-distribution or from airport choice to mode 
choice and further on to generation-distribution, one links the modelling 
steps by the quasi-direct format using the representative utility ftmction of 
the lower level models in the upper ones as an additional explanatory 
variable, which we call modal utility index U. 

In addition at the level of the discrete choice models the explanatory 
impedance variables used in the model specifications are computed 
considering the probabilities of the lower level model as weights. The idea of 
linking the models in the forecasting system is shown in Figure 3. 

Applying this system approach, a consistent simulation instrument can be 
constructed which reflects the impacts of supply changes through all 
instances at any level. The effects of supply changes at an airport (e.g. a new 
0-D service, increasing aircraft fees, low cost tariffs) can be analysed in 
detail. No matter whether these are intramodal impacts, say the competition 
of airports about market shares as well as the competition of different levels 
of service (non-stop versus via connections), or multimodal impacts, say the 



Airport Choice & Competition - a Strategic Approach \Q 

substitution of air traffic by inter city high-speed trains or car but also the 
vice versa case is possible due to the offers of low cost carriers, or 
intermodal impacts, say the co-operation of air and rail services on the 
access/egress side to/from the airport (like a lot of tourist companies already 
include in their price offers in form of rail & fly tickets). Therefore the 
interdependency of airport choice and travel demand can be analysed out of 
different point of views. 



Generation/Distribution 



Modal split utilitr Index 



Airport Choice i;^-"-^' 



E 
Z 
© 




Impedances 



Access/egress Choice ^^';^' 




_, . , . utility Index, 

Slot choice Impedinces , 



Fig. 3. Linking the models in the forecasting system 

Instead of extending this paper by the theory of all modelling steps used 
in the system approach it is referred to various publications. A detailed 
theoretical background of the modelling steps is given by the following 
literature. 

For the generation-distribution modelling it is referred to Gaudry- 
Mandel-Rothengatter 1994a, 1994b, Sen-Smith 1995 and Last 1997 and for 
the mode choice modelling see the publications of Mandel 1992, Mandel- 
Gaudry-Rothengatter 1991-1994-1997 and Mandel et al. 1997. The more 
general focus of discrete choice modelling you'll fmd in Domencich- 
McFadden 1975, Manski-McFadden 1981 and Ben-Akiva-Lerman 1985. 
The quasi-direct format is explained by Tran-Gaudry 1994. Concerning the 
assignment procedures there is a lot of literature therefore it is referred to a 
more general operations research summary by Neumann 1974 and a more 
transport oriented publication by Gallo-Pallottino-Florian 1984. For details 
of the assignment procedure used in the system approach see Last-Mandel 
1997. As introduction an overview concerning all steps is given by Ortuzar- 
Willumsen 1990. Specific information concerning air transport is found in 
Doganis 1991-1992 and in several publications of the air industry and their 
associations like ICAO, I AT A, CAC and EC AC or national ones like ADV 
in Germany. 



Airport Choice & Competition - a Strategic Approach 1 1 

After having shed a Hght on the global approach we are going into some 
modelling details which are of relevance for a sound and detailed analysis of 
interdependencies in the air transport market as shown e. g. in the examples 
of the final section. 

Therefore we'll highlight the modelling aspects concerning two issues. 

1 . Non-linearity: to enable the models to capture existing thresholds where 
consumers strongly react due to changes at the supply side caused by e.g. 
infrastructure investments, pricing strategies or service changes. 

2. Notions of competition: to explain the interdependencies of different 
modelling steps as well as ways to measure it so that the consumers' 
reaction on supply changes can be quantified and finally evaluated. 

For a detailed discussion we'll refer to the papers (Mandel 1999) 'The 
Interdependency of Airport Choice and Travel Demand' and 'Measuring 
Competition in Air Transport'. In the following we'll withdraw parts of the 
text. 



Airport Choice & Competition - a Strategic Approach 12 

3. NON-LINEARITY AND THE CONSUMER'S CHOICE 

On most flight journeys consumer have the opportunity to start their trip 
from more than just one single airport. Just in Germany there are 16 
international airports and several regional airports so that often a situation 
appears where the consumer can choose out of a whole bunch of alternatives 
all serving his needs in nearly the same way. It could happen that more than 
four airports offer the same destination and in addition at each airport non- 
stop as well as via services are available. After the consumer evaluated the 
different opportunities, he selects only one out of the available set of 
alternatives. This is a classical discrete choice problem. For details see the 
suggested literature about mode choice and discrete choice modelling in 
section 2. 

Here the focus is on the differences between the chosen and the standard 
approach and the resulting advantages. To understand the issues in an easy 
way we will also refer to examples taken from the field of mode choice. 

3.1 Properties of the linear standard model 

The "classical" linear Logit model specification normally assumes 
(Gaudry 1992): 
(i) linearity in variables; 
(ii) the exclusion of characteristics of other alternatives j e C„ from the 

representative utility of the i-th one (/ e C„, i^j); 
(iii) equal "abstract" or "generic" coefficients for the network charac- 
teristics, a constraint that is not necessary but is frequently imposed. 
These assumptions lead to unrealistic properties. Because of (ii), the 
standard model implies: 

a) equal cross elasticities of demand: this means that setting up a bicycle 
path between two cities will draw the same percentage of travellers from 
the plane, car and train or in the sense of airport choice the same 
percentage of travellers will be drawn from all considered air services. 
Furthermore (iii) implies identical values of time across the alternatives: 
this means that representative train and plane users (mode choice) or 
non-stop and via flight passengers (airport choice) value time 
identically; 

b) the exclusion of complementarity among alternatives; 

c) that only differences in the level of characteristics matter, or that the 
function is not homogenous of degree 0: in consequence doubling all 
fares and income will change the market shares. 

Because of (i) the standard model further implies that: 



Airport Choice & Competition - a Strategic Approach 



13 



d) the effect of a given difference in transport conditions is independent of 
the service level characteristics so that the response curve to changes in 
service characteristics is symmetric with respect to its inflection point 
(see figure 4). For instance, a 30 min rail travel time reduction has the 
same impact on choice probabilities for the Hamburg-Hanover (180 km) 
origin-destination pair as for the Hamburg-Munich (823 km) pair. 
Similarly increasing the air tariffs by 50 DM (US$ 28) has the same 
impact on choice probabilities for the Hamburg-Frankflirt (540 km) 
origin-destination pair as for the Hamburg-Paris (1060 km) pair. Further 
on adding an amount of 20 DM (US$ 11) to the price of travelling by 
plane will have the same impact as adding 20 DM to the price of 
travelling by train. The same holds if one directs this example to non- 
stop and via-altematives in the aviation sector. 

Generally speaking, symmetry, with respect to the inflection point, 
implies that potential asymmetry of behaviour, where consumers/travellers 
suddenly start to react and then change their behaviour, can not be detected; 




range of thresholds 



H -X 



60 



75 



decreasing travel time 
90 in minutes 



— Linear Logit 
• Box-Cox Logit 




Fig. 4. Linear Logit versus Box-Cox Logit 

e) coefficients for the constants and for the variables common to all 
altematives are underidentified, which means that, for these variables, 
only differences with respect to an arbitrarily chosen reference can be 
identified. 

We also note in passing that the logit form requires that 



Airport Choice & Competition - a Strategic Approach 14 

f) the choice probabilities go to zero (one) when the representative utility 
Vi goes to -co (+co) so that (see figure 4) one cannot model thick tails due 
to specification error, modeller ignorance, compulsive consumption or 
captivity to alternatives. The latter case includes the situation of business 
travellers which have to do a one day return trip and are therefore not 
elastic to price. 

3.2 The Box-Cox device 

To bypass most of these constraints (generally speaking, only (c) and (f) 
will remain), the Box-Cox transformation is used: 



X^=\ 



jx^. - 1) 



^v (1) 

In^,, if A,=0. 



Hence, the choice model based on the logit function can be written as: 

exp {fi:X, + X /„X<^') 

^(0. = '-^ , (2) 

X exp (fi,X , + 2: fi,X'^,'') 

\v]\txtXi=Xj - 1 are regression constants. 

If /Ijt,- is equal to 1 (or zero), then the variable is entered in its linear (or 
logarithmic) form. Since the transformation is continuous for all possible 
values of the /^-parameter, but defined only for a positive variable, it is 
clearly understood that in above formulas some of the X^„'s cannot be 
transformed: the constant, the dummies and the ordinary variables that 
contain negative observations. Variables that contain positive and null values 
can be transformed as long as a compensating dummy variable is created 
(Gaudry et al. 1993). 

3.3 Visual and economic significance 

Figure 4 clearly shows the difference between the linearity and non- 
linearity of a variable. The asymmetric curve (in respect to its inflexion 
point) given by Box-Cox transformation (Box et al. 1964) of the strictly 
positive variable X^ illustrates the error which will occur when a non-linear 



Airport Choice & Competition - a Strategic Approach 15 

variable is forced to be linear. For example, assume X^ denotes total travel 
time: in the linear case, the value A', equal to 30 is associated to the 
probability P equal to 0.25; in the non-linear case, the probability is higher if 
A < 1 and smaller if -^ > 1 . Hence, if one forces a non-linear variable - or in 
equality the utility function - to be linear, this will result in an over- or 
underestimation of the probability related to this variable. In addition to 
asymmetry of the response function (i t^ 1), reaction thresholds can be 
identified. 

The Box-Cox transformation of the strictly positive variables of the linear 
Logit model leads to the Box-Cox Logit model with an asymmetry of 
response, as shown in figure 4, because the effect of a unit change in the 
service will depend on the levels of the variables Z^y for all values of A^j not 
equal 1. This can be seen by examining the partial derivatives of the 
representative utility function Vj of they-th mode. It is obvious that the effect 
of additional service will be smaller at higher service levels than at lower 
ones if A;,j is smaller than 1. These diminishing returns mean that given 
absolute reductions in total travel time have more impact when total travel 
times are low than high: a gain in travel time of 15 minutes means less on a 
long trip than on a short one. The same effects appear in the case of an 
increase of air tariffs. Conversely, increasing returns exist if i^^- is larger than 1 . 

Clearly, if one is considering very small changes in the service levels of a 
mode, the mathematical form used does not matter very much because one is 
forecasting in the immediate neighbourhood of current sample values. 
However, if one is considering significant changes in service levels, such as 
increasing aircraft fees or decreasing air tariffs by a third or reducing train 
travel time by one half with high-speed trains, then curvature is decisively. 

3.4 Asymmetry 

To illustrate the asymmetry of the response functions due to the inflexion 
point of the curve and the threshold effect mentioned before, figure 4 shows 
a general example of a response curve for an alternative with respect to the 
variable travel time while all other conditions (characteristics and 
alternatives) remain unchanged. On the x-axis the change in travel time is 
displayed (t minutes decrease in travel time on the air service alternative) 
and on the y-axis the change of the probability choosing this alternative is 
given. Hence the interdependency of airport choice and travel demand is 
obvious when the probability is multiplied with the total demand of the 
origin-destination pair which will show you the demand for the alternative. 

To describe asymmetry more formally one first has to define the 
inflexion point of the curve. At this point the curvature changes its functional 
shape from convex to concave and one can compute the value of the 



Airport Choice & Competition - a Strategic Approach 16 

inflexion point {Pt^ , /J by equating to zero the second derivative of the 
alternative share in respect to the travel time. The response curve can be 
called asymmetric with respect to its reflection point if equidistant reductions and 
increments of travel time /„ by A/ [that is,^ = t„- At and t; =t„+At] will give 
different absolute values, namely AP*= \P,^ - P.\ and AF=t^,^-P.\ : otherwise 
the curve has to be called symmetric. More formally, one can define 
asymmetry, as in Laferriere and Gaudry (1993) in terms of the partial 
correlation ^ of P, and (1-P.): this yields an indicator that is necessarily 
between and 1. 

Threshold. A threshold effect occurs when the travel time reaches a 
critical value oft beyond which any further reduction of / to /' =t + At provokes a 
more substantial growth of the mode share P, than an equidistant increment 
of no /"= / - A /, so that the absolute difference of the mode shares \P,—P.\ is 
higher than I P^-Pj. ' 

The word threshold implicitly involves an individual evaluation of the 
perception of change; hence it is up to decision maker to define his threshold 
by exploring the percentage of alternative share increment which he will 
consider as a threshold i. e. which will satisfy his opinion about a threshold. 
More formally a critical value 7 has to be defined so that the absolute 
difference of \P, -Pj = U+tj) I/^ - P J . Alternatively, AP. = {I+tj) AP,and 
hence 7j = {AP^ /AP,)-l. From a visual point of view, one would intuitively 
expect to find the thresholds to be in the range given by the grey zone in 
figure 4, where a reduction of one unit would increase the probability of 
choosing the alternative by an additional 20% (7 = 0.3), so thaPA would 
be equal to 6 timesR\ 

It is obvious that the given results in figure 4 are based on a ceteris 
paribus assumption: consequently a variation of other mode specific 
characteristics like frequency and travel cost would imply a change of the 
location of the response curve so that the threshold would have to be 
relocated. 

It has to be mentioned that in general by interpreting the results shown in 
figure 4 one has to take into account that the travel time represents the time 
of a door to door trip. Therefore a change of the access/egress services can 
have an important impact on the choice of an alternative. 

3.5 Other considerations 

The purpose of the latter example is to visualise the asymmetry of the 
response functions, the existence of the thresholds and the impact of travel 
distance on consumer behaviour: it is clear that for a detailed analysis of an 
investment or planned action it would of course be necessary to consider in 



Airport Choice & Competition - a Strategic Approach 17 

addition the impact of travel cost, frequency, access/egress characteristics, 
etc. 

The examination shown in figure 4 also can be done in reverse direction 
where one first defines the probability of choosing the mode and then 
computes the necessary characteristics which satisfy this condition. Different 
kinds of services, which are related to different actions can be represented by 
changes in the underlying characteristics. Implicitly there is the possibility to 
verify the optimal investment by relating it to the alternative specific 
characteristics that maximise revenue. 



Airport Choice & Competition - a Strategic Approach 18 

4. COMPETITION CONCERNING AIR TRANSPORT 

Before we go into some formal details we first have to state which forms of 
competition appear and how they can be explained/ For this purpose it is 
necessary to view the air market from the outside to identify all structural 
components of competition. The following different types of competition in 
the air market can be distinguished: 

- competition on a single route, 

- competition between networks, 

- competition for infrastructure and 

- competition between access/egress points. 

The classical notion of competition is the rivalry between two carriers on 
a certain route. This kind is usually expressed in market shares kept by the 
competitors. The second one is measured in more aggregated market shares 
and means the rivalry between two and more airlines as well as those 
between airline alliances. 

Competition for infrastructure comprises e.g. the fight for slots or 
ground-handling capacities. In this case limited resources on airports' are the 
reason for the conflict between the airlines. An often misinterpreted form of 
competition is that between airports, mostly owing to a special view of the 
air sector. 

Out of a general transport point of view the supply side should be a result 
of the offered O-D services which include the airports and land-based 
access/egress modes while the demand side is given by the travellers with all 
their needs and priorities for a trip. 

An airports' attractiveness in a condensed air market depends strongly on 
the capacities, the pricing structure, the land-sided accessibility and the non- 
aviation supply. Those factors and the carrier-related supply based on them, 
all together will finally attract customers, i.e. travellers and also shoppers. 
Hence, besides infi-astructural matters the potential of customers is the major 
driving force for airlines to choose an airport. 

The competition of airports on the cost side (passenger charges, landing 
fees, ground service) is already ongoing especially if one considers the non- 
harmonised airport costs in context of the air alliance network optimisation. 
Airports with 'bad' price structures, i.e. high costs for airlines (passed to the 
travellers through the fares), are already facing the problem that clients - 



Naturally modellers prefer to explain the interdependencies instead of just describing 
observed situations because their aim is to understand the underlying structure of a system. 

Due to the fact that at least Uvo airports are required for air routes, the constraints must not 
necessarily exist on both airports because slots on unconstrained airports must fit to their 
counterparts on the constrained ones. 



Airport Choice & Competition - a Strategic Approach 19 

travellers and airlines - look for alternatives. Therefore airlines try to 
combine their network synergies at 'better' airports. 

While hubs are already working at this problem other airports still 
hesitate and focus on 0-D market services. That is the reason why they 
estimate the influence of varied fees and charges on carrier decisions as 
relatively small in contrary to routing optimisation. But airlines want to 
satisfy the consumers needs as profitable as possible and consumers' try to 
obtain their optimum from the supply offered. 

Neglecting such obvious dependencies and standing at the side, waiting 
what airlines and travellers do, is certainly an unconventional strategy which 
might be applicable if the airport is in a monopolistic situation e.g. due to 
bilateral agreements but will not suit those airport managers that view 
travellers as their clients. 

They will agree to the idea that persons intending to travel from an origin 
to a certain destination have to be convinced to choose a route via their 
airport. But this route competes with those through other airports and routes 
that use only ground-based transport modes like (high speed) trains or 
private cars. Hence, airports are in a very large competition that should be 
considered as completely as possible in the decision-making processes. 

Taking into account only the air transport system, as airlines often do, is 
not sufficient, when one aims at the traveller as the final driving force. But 
more crucial is the scope of airports when analysing the market. Neglecting 
neighbouring airports and also those further away as competitors falsifies 
any serious evaluation and in consequence any planning. 

The planning and analysis instrument must therefore cover all main 
modes - road, rail and air for the multi-modal competition. Further, the 
corresponding access/egress systems have to be considered for inter- 
modality beside detailed representations of the air transport system itself to 
assess the intra-modal rivalry between airports and/or carriers. 

Well-developed planning instruments based on a system approach 
simulates the complete supply side a travelling individual* is confronted with 
and from which it has to choose its path from the origin to a desired 
destination. Interpreting the different paths as alternatives in a choice 
process, competition could be measured in terms of the various probabilities 
to select one of the possible paths. 

It is important to note that the traveller has to take a discrete decision 
about the alternative to be used because he can only use one alternative at 
each time. The choice among the set of available alternatives depends on 
subjective preferences and/or on the alternatives' characteristics. Neglecting 
individual preferences for the moment, the traveller compares the 



* We are focussing here on individuals who have just decided for a trip and do not cover those 
which are still in the process of decision whether to travel at all. 



Airport Choice & Competition — a Strategic Approach 20 

alternatives on the basis of their measurable characteristics like e.g. travel 
cost, travel time, comfort and security. 

The preferences come into the decision process when the travelling 
person weights the 'objective' characteristics for each alternative due to its 
personal rating. In economics the resulting measure is referred as 'utility', 
i.e. the satisfaction one receives from choosing one alternative. It is obvious 
that consumers evaluation varies between different individuals which for 
example can be segmented according to their socio-economic situation 
(income, age, gender) or their trip purpose (business, vacation, visiting 
friends and relatives). More formally one can speak here of the consumer's 
'elasticity' in respect to any alternative's characteristic. 

The elasticity tj just measures the ratio of the percentage variation of a 
dependent variable Y due to the percentage change of an independent 
variable Xk(k € (1, ....AT^y given all other independent variables fixed at their 
observed value. As dependent variable one can use e.g. the total flow or 
market share - 7) or /),■ - and assess the impact of passenger fare as an 
independent variable, keeping all other independent variables like travel 
time, frequency, service attributes, etc. unchanged. 

The more general form for any elasticity is for the point measure is: 

^0-..,; = !^^ (3) 

As we want to focus on individual aspects of consumer behaviour we 
choose a disaggregate probability approach P(i)„ with underiying Logit 
function (see section 3). Now the direct elasticity 7 of the probability of a 
consumer n choosing alternative i with respect to a change in the 
characteristic Xkm is given by: 

In addition it is interesting to know how the changes of a characteristic 
xiy„ effects the probability P(i)n. Therefore the cross elasticity of the selected 
probability alternative i with respect to a variable of alternative J can be 
computed. But in addition we want to bypass some problems caused by the 
properties of the classical standard Logit approach and all the interpretative 
problems such as equal cross elasticity's (see section 3). Equal cross 
elasticity's of demand imply identical 'values of time' across all alternatives: 
this means e.g. that non-stop and via flight passengers value time identically. 
Therefore a non-linear transformation to strictly positive variables is applied 
like stated in section 3 (see also Box-Cox (1964)). 



Airport Choice & Competition - a Strategic Approach 2 1 

For the interpretation of elasticities one should be aware that the values 
given are computed on the basis of a 1% change of the variable Xk. As the 
formulas (4) and (5) show, the result is a share value and therefore the 
interpretation always has to be put in context to the demand an alternative 
attracts. If the elasticity is small but the attracted demand of that alternative 
is high the demand effects could be bigger than in the reverse case. More 
formally the own elasticity is: 

r] = /3j^x^^[l-P(i)] (5) 

where the elasticity increases with ^ and falls with the level of P(i). 

As we noted in the beginning of this section there are various areas of 
competition which have to be considered in total to reflect consumer 
behaviour. Therefore based on the shown principle idea of elasticity one can 
construct an equation which considers these elements. Instead of using just 
one discrete choice part in the formula we can add the interesting fields of 
airport choice, access/egress choice, time slice choice and airline choice in 
the following way (with simplified notion). 

(77 of alternative) = (7 of total flow) + (7 of mode) + (7 of airport) + 

(77 of access/egress) + (7 of time shce) + 
(7 of airline) (6) 

This expression allows us to compute the elasticity of demand of an 
altemative with respect to any variable Xk considering the impacts on the 
following types of competitive situations: 

- competition of destinations (substitution, complementarity) 

- competition between the modes (air, rail and road), 

- competition between co-operating modes (air-rail, air-road), 

- competition between air services at airports, 

- competition of access/egress modes to/from the airports, 

- competition for time slices at airports and 

- competition of airlines. 

As already mentioned above this formula should be used for any market 
segment, for example business travellers tend to have a lower elasticity 
concerning travel expenses than for travel time and the reverse holds for 
holidazmakers. 

Obviously to calculate such complex elasticity structures which allow 
detailed analyses at any point, a system approach is needed. It has to be 
assured that the interdependency of different models is reflected properly 
and therefor models have to be linked so that the results are consistent. One 
way of doing this is using the 'quasi-direct format' where the different 



Airport Choice &. Competition - a Strategic Approach 22 

model steps are linked by the representative utility function of preceding 
models in the subsequent ones as additional explanatory variable. This 
approach can be enriched by considering the probabilities of the preceding 
models as weights when the explanatory impedance variables in subsequent 
models are computed (see Mandel (1999)). 

As experts expect that the main focus of interest will be on airport 
competition in the coming years, we will present some strategic analyses 
examples in the last section based on a restricted sequence of elasticities as 
shown in equation (7). 

•{Ti. Xk) = '{T. Xk) + iP(mode)„, Xk) + •{P(airport)n, Xk) + 
■{P(access/egress)n, Xk) (7) 
Finally it has to be stated that most of the strategic analyses shown in the 
following section are aggregated concerning destinations and trip purposes 
and that the elasticity's used are documented in Mandel et al (1994) Gaudry 
et al (1994 a) and Mandel (1999). 



Airport Choice & Competition - a Strategic Approach 23 

5. SELECTED STRATEGIC ANALYSES 

To show the effects based on the approach outlined in the sections above 
we are going to present the reader results on the microeconomic level (of 
individual consumers) as well as on the macroeconomic level because the 
evaluation of strategic scenarios has often to be done out of a global point of 
view. Therefore all simulations are computed regarding the specific 
consumers' elasticity but most of the results are aggregated and displayed on 
the macroeconomic level. Despite the macroeconomic orientation of 
strategic decisions where we focus on situations of competition and the 
related consumer behaviour which can be influenced by strategic means, we 
will show the underlying elasticity or response curves to clarify the 
theoretical background. 

For the sake of clarity, the results displayed will be restricted to the 
geographic shape of Germany although consumer behaviour beyond the 
German border is affected and of course considered in the computations. 
Within this section some possibilities of strategic scenarios are shown. All of 
them are focusing on consumers' reaction to supply changes. Thus the 
results are reflecting the elasticities of consumers. 

The first analyses will deal with the market situation showing the 
different point of views, from the airport and from the region / consumer. 
The second part will refer to the interdependencies of air transport and the 
total transport sector by analysing the problem of airport location and the 
access/egress systems. The last part will present some simulation scenarios 
and analyse basic consumer behaviour to ceteris paribus supply variations. 

In all scenarios the results reflect the interaction of multi-, inter- and 
intramodal effects due to the applied system approach stated in section 2. 
Obviously other strategic scenarios can be simulated and evaluated in the 
same way or even in a more detailed manner depending on the client's 
needs. 

5.1. Market situation 

Here we distinguish between two different point of views. Focussing a 
certain airport because we might be in charge of managing it or regarding the 
supply situation from a regional scope because we are planning to make a 
journey or we are e.g. responsible for the regional accessibility of major 
transport infrastructure. 

Both locations are interesting because at the one hand the airport gets a 
feeling of his market position and on the other hand decision makers will 
understand more about consumers choice due to the displayed competing 
alternatives. 



Airport Choice & Competition - a Strategic Approach 



24 



5.1.1. Catchment area 

If the scope of analyses is those of an air transport service supplier, we 
might be interested in the question where our customers come from. If we 
ask e.g. for the market dominance of the Frankfurt Airport in Germany, we 
obtain the catchment-areas by aggregation over regional and trip purpose 
specific transport flows using the considered airport. The resulting figures 
show the realised market shares for this airport. 

Figure 5 reflects the sphere of influence according to the intercontinental 
market. Frankfurt as Germany's major hub is offering a large number of 
long-haul connections, so its market dominance covers a larger area in this 
market section than in total. Other international airports as Hamburg, 
Munich, Stuttgart or Dusseldorf are able to claim significant market shares 
in the domestic and charter segment as well as towards selected destinations 
abroad. 



Mani«4 aturvt m % 

■ to % b» too % 

■ •0% mSO % 

■ ro % ta w « 
Q M % M ro % 

y to % IB 20 % 
Q 1 % M 10 % 

D OHta 1 % 




UnwnbCH^g: 5 % 



LtfTMM {'Fr Y 19 % 



AlftaM(Fr.). a% 



'f INN 



Fig. 5. Market shares of Frankfurt 1991: intercontinental destinations 



It is obvious that the hinterland of an airport cannot be described by one 



Airport Choice & Competition - a Strategic Approach 



25 



or several concentric circles. The shape rather depends on specific 
characteristics of the airport and its competitors, like number of destinations 
and flights offered or the accessibility by earthbound feeder systems. So the 
catchment area (to all destinations and with more than 10% share) of 
Frankfurt extends to 600 km in the north-south direction, while in east-west 
direction only to 300 km. Of course the catchment area exceeds the German 
borders but the main access lines by the land based modes are in north-south 
direction. For the sake of understanding all figures are restricted to the shape 
of the German borders. 

Referring to the last subsection, you will find in the scenario 'route 
inauguration' the catchment area of Frankfurt for the North American 
Market (figure 18). Comparing these catchment areas it is obvious that they 
vary according to the market, as they are an endogenous result of the system 
approach which takes into account the market specific competition of 
airports / air services. 



Tld. 


puMngan 


■ ZMUUO 


■ icoiono 


□ 


90 B too 


C 


2Sl0 90 




lOU 23 


" 


iM 10 




010 S 




+ LNZ 



+ INH 



Fig. 6. Passengers at Frankfurt by origin for intercontinental destinations 

1991 



Airport Choice & Competition - a Strategic Approach 26 

If we substitute the regional market shares in figure 5 with the absolute 
number of passengers the spatial demand pattern for Frankfurt Airport gives 
deeper insight in the market potentials (figure 6). 

Beside the extended area of Frankfurt and the densely populated counties 
south of it, where most passengers are originating, a remarkable number of 
people are withdrawn from other metropolitan areas in Germany, although 
there are international airports in those counties (e. g. Cologne, Hanover). 

In the next figure we compare the different spheres of influence of four 
German airports situated quite close together in the high populated region 
"Ruhrgebiet". Market shares shown are a forecast for the year 2004 to/from 
all destinations. The first two airports Cologne (CGN) and Dusseldorf (DUS) 
offer a huge number of destinations served by non-stop-flights, not only 
inside Europe, but also to some intercontinental airports. In addition almost 
every holiday destination throughout Europe as well as some of them in 
other continents are connected to Dusseldorf and Cologne by non-stop or at 
least direct flights. The other two airports Munster/Osnabrueck (FMO) and 
Paderbom/Lippstadt (PAD), offer only a few international destinations to 
selected hubs like London and Amsterdam. They focus on domestic air 
transport and charter flights for holiday trips to destinations around the 
Mediterranean Sea. 

The market shares in the counties situated quite close to the airports are a 
result of the available destinations offered. In case of Dusseldorf and 
Cologne shares reached go up to a maximum of 80%, while 
Munster/Osnabrueck and Paderbom/Lippstadt can realise only values up to 
60% resp. 40%. 

In addition, Cologne and Dusseldorf can be reached by urban mass transit 
as well as by high speed trains, while the level of service in public transport 
to Munster/Osnabrueck and especially to Paderbora is quite poor. Together 
with the lack of destinations compared with DUS and CGN, this influences 
the total size of the catchment area of an airport. So both, Dusseldorf and 
Cologne have market shares of almost 40% in the Paderbom area and still 
almost 20% in the Munster area, while in comparison to that 
Munster/Osnabrueck and Paderbom/Lippstadt do not gather any passengers 
from the Cologne or the Dusseldorf county. 

Finally, from this catchment areas, one can see, how hard different 
airports compete. In case of the Dusseldorf airport, market shares decreases 
by more than 30% between adjoining counties towards the airport of 
Cologne. The same holds for Cologne's sphere of influence towards DUS, 
while to other directions the specific market shares decrease more moderate. 



Airport Choice & Competition - a Strategic Approach 



27 



Cologne-Bonn 

■ 40 % toflO % 
B 40 % u 40 % 
O 20 %ia40-( 

Q '0 Mo :o % 

8t %io 10 s 
»«IOw t % 



Dusseidorf 


■ wxtoao% 


B^ MbOOX 


DM%ta*OH 


Q 10 X IB 20% 


Q 1 %!• to% 


G tMlOo 1 % 




Cologne-Bonn (CGN) 



Dusseldorf (DUS) 





Munster-Osnabruck (FMO) 



Paderborn-Lippstadt (PAD) 



Fig. 7. Market shares at four different Airports; forecast 2004 



Airport Choice & Competition - a Strategic Approach 



28 



5.1.2. Regional differences 



Coming back to the question why airport choice has to be considered a 
small example (from 1991) will demonstrate the choice problem a consumer 
faces and which even become more difficult in a liberalised air market with 
capacity constraints at different airports. 

Comparing two German business travellers - e.g. one living in famous 
Heidelberg (county) and the other one in the neighbouring county of 
Karlsruhe - their decisions concerning the chosen airport for a trip will vary. 
If we neglect possible individual preferences a set of extemal factors 
influences their choice between possible starting points for a flight. If we 
study trips destined for e.g. North America (see fig. 8) we will find that the 
probability to travel via Frankfurt-Main Airport for both travellers is around 
ninety per cent. This is not very surprising due to the fact that Frankfurt 
dominates the German market as the largest hub and being the homebase of 
the national carrier Lufthansa. 



%: 




(Dus s eldorf ^ i 
\^ 



Heidelberg 




(Basal ) 



Karlsruhe 



Fig. 8. Airport choice, destination: North America, trip purpose: business 

Much more interesting are e.g. the probabilities to choose alternative 
airport for business trips crossing the Atlantic. As described in the sections 
above the choice of airports is determined by a set of factors including 
accessibility, offered frequencies and destinations. The figures above shows 
how the combination of these factors influences the probability to choose 
one of the remaining alternatives for such a tnp. While the airport 
characteristics are equal in both cases the accessibility by private and public 



Airport Choice & Competition - a Strategic Approach 



29 



transport differs. In general it was found that business travellers prefer 
strongly the airport, which offers the highest flight frequencies and the 
shortest duration of the whole trip, including access and egress. 

When analysing vacation trips one observes a completely different choice 
structure for the travellers in the example above. Due to lower restrictions in 
time but higher price sensitivity holiday-makers prefer the most convenient 
kiss&fly-access to possible airports offering non-stop or via flights to start 
their journeys but they are also open to choose other alternatives as long as 
the price differs significantly. Figure 9 depicts the choice probabilities for 
holiday travellers from Karlsruhe and Heidelberg, respectively, to an Italian 
destination. 




Heidelberg 




Karlsruhe 



Fig. 9. Airport choice, destination: Rome (Italy), trip purpose: vacation 

Major differences rise from other characteristics, namely the distance 
between the origin and the airport or the availability of a non-stop flight 
(even if only once a week) or the accessibility by public transport. So 
travellers from Karlsruhe prefer strongly Stuttgart Airport. Frankfurt, which 
is situated additional 50 km away, can only attract a market share of 13%, 
although much more flights are offered than in Stuttgart. Due to a missing 
non-stop-flight to Rome, only 8% remain for Strasbourg Airport, despite the 
fact that is it is the nearest one to the area of Karlsruhe county. 

When starting a holiday trip from Heidelberg, Frankfurt is certainly the 
best choice for two third of all vacation travellers. But the second best 
alternative via Stuttgart gets still 30%, while choosing other airports as e.g. 
Strasbourg or Basel will be an exception. 



Airport Choice & Competition - a Strategic Approach 



30 



For the sake of understanding only this small example of Heidelberg and 
Karlsruhe was presented. By the way, in the meantime (1997) in the area of 
Karlsruhe a regional airport (Baden-Airport) opened and already offers 
interesting services to tourist centres which unfortunately could not be taken 
into account for these examples based on 1991. Of course such examples can 
be extended when one moves to areas where a lot of services and airports are 
competing, like in the Rhein-Ruhr, Berlin, Paris or London area, and one 
analyses all possible destinations and alternatives. Some analyses follow 
below 

After we saw, how people from to different counties chose their airport, 
we now depict on a single area and compare differences in airport choice 
depending on the passengers destination. Figure 10 shows the market shares 
for some destinations, different airports can realise in the Hanover area. Here 
consumers airport choice is shown for the trip purpose "vacation". Analysis 
can be done as well for other trip purposes, destinations or regions. 



100% 




Mddto-. South Arrerica. North America 

CarnbeanSea 



Africa wihout 
MediWf ranean S«a 



Far East. Austrata 



Balaarian klandi 



Fig. 10. Airport choice from the county of Hanover to different destinations 

1991 

This airport choice differs, strictly depending on 

- availability of non-stop-flights to the specific destination at the airports, 

- distance (travel-time) between the county and the different airports, 

- price for the flight to the specific destination from the airport and the 
costs of access/egress, 

- total travel-time (access/egress, check-in/out, flight-time) 

- frequency offered on the specific routes offered at the airports. 



Airport Choice & Competition - a Strategic Approach 



31 



So as an example people from the area of Hanover chose Hanover 
Airport for holiday-trips to the Balearian Islands with a probability of more 
than 70%, due to access/egress plays a major role on such a short-haul flight. 
In addition non-stop-flights to that destination are offered at Hanover Airport 
with high frequency. 

On the other hand passengers from that area travelling towards North 
America prefer the Frankfurt Airport (48%), as the prices offered there to 
that destination in average bet the costs from starting at Hanover, including 
the railway fare when using the available high-speed trains Hanover - 
Frankfurt. Additionally in this case the consumers more accept an exceed of 
time for access/egress to Frankfurt, due to the longer total travel-time on 
such long-haul trips. 

The next possibility to analyse the regional / consumers' point of view 
concerns the competing alternatives considering the different transfer 
locations. This analyses is based on the principle idea shown in the 
beginning but allows in addition to show hub-potentials, if an aggregation 
over all regions to one destination is computed. Anyhow to understand the 
principle one example is selected which shows the competitive situation 
from Bielefeld to Hong Kong in 1994. By the way Bielefeld is located south- 
west of Hanover where the EXPO 2000 will take place. 




Fig. 11. Airport choice from Bielefeld to Hong Kong, full fare business class, 

1994 



Airport Choice & Competition - a Strategic Approach 



32 



The influence of pricing strategies (which will be shown in the last 
subsection of this paper) upon the hub potential is obvious if one compares 
Amsterdam and Hong Kong, whereby the situation of the latter changed due 
to the new airport and the much higher fees / charges. 

From Bielefeld towards the destination of Hong Kong, full fare business 
class OrrP)-passengers in majority prefer flights from the two nearest 
available airports (Hanover-HAJ and Munster/Osnabrueck-FMO) although 
no non-stop or direct flights are offered there. Another third of all those 
passengers take the route via Frankfurt (FRA), as it is equipped with non- 
stop flights to Hong Kong and Frankfurt can be reached in almost adequate 
time by rail or road. Amsterdam (AMS) can still realise a market share of 
more than 5%, although it is quite far away, but can be reached by rail in a 
good way and offers non-stop-flights as well as Frankfurt. The rest of the 
shown alternatives are only rarely chosen, as they neither offer non-stop 
flights nor those airports are situated very close to Bielefeld. 

Table 1. Airport choice from Bielefeld to Hong Kong, full fare business 
class. 1994 



Alternatives 


origin from 


via 1 


via 2 


destination 


share for full 
fare business 


1 


FRA 


— 


— 


HKG 


34.5 


2 


AMS 


— 


— 


HKG 


5.4 


3 


HAJ 


CPH 


— 


HKG 


3.3 


4 


BRE 


CPH 


— 


HKG 


0.6 


5 


FMO 


FRA 


— 


HKG 


28.5 


6 


HAM 


CPH 


— 


HKG 


0.9 


7 


DUS 


CPH 


— 


HKG 


1.2 


8 


HAJ 


ZRH 


— 


HKG 


3.0 


9 


DUS 


ZRH 





HKG 


2.4 


10 


HAJ 


FRA 


— 


HKG 


20.1 



In table 1, we can focus on the stop-over connections and the hubs people 
change plains. Here the consumer prefers flights via Frankfurt, offering good 
connecting times, due to the high frequency of the feeder flights from FMO 
and HAJ to that hub. This two alternatives already form 48.6% of the 54.9% 
passengers using FMO and HAJ at all. Other hubs chosen are Zurich (ZRH) 
and Copenhagen (CPH). As we regard the full fare market sector, other big 
European hubs like London or Paris do not play any role, due to their 
geographic situation according to the Germany - Hong Kong routes. It has to 
be remarked that no alternative with two stopovers has been selected and we 
just display ten alternatives which summed up to 100% although there are 
plenty more possibilities. It is also obvious that the same analyses for 
economy class will show another preference of altematives, e.g. share 
distribution. 



Airport Choice & Competition - a Strategic Approach 
5.2. Interdependencies 



33 



This subsection will show the interdependencies of air transport with the 
total transport sector. It will be obviously that decision makers have to 
consider the overall framework to know all influencing factors to minimise 
the risk in strategic planning. Based on an existing case study, choosing a 
location for an airport, we analyse the total passenger demand, modal spilt 
and access/egress effects. Finally two examples will analyse the public 
access/egress mode. 

5.2.1. Location of an airport 

Mode choice, airport choice and access/egress choice as a part of a traffic 
forecast can help to come to a decision where to place a new airport best. 
The comparison of five locations for a new airport near Berlin on several 
points of view will be shown in the following figures. 

A set of locations has been evaluated with respect to different measures. 
All considered alternatives are located in the south respectively south-west 
of the German capital. The corresponding scenarios cover beneath the single 
airports also two airport systems that are combinations of Tegel and 
Schoenefeld as well as Jueterbog-W. and Schoenefeld. The figure 12 depicts 
the five different locations considered. 




/ ^ |]uetert)og-"o7| \ ^ ® 

Fig. 12. Comparing locations for an international airport for Berlin 

Various indicators could be assessed to evaluate the relevance of certain 
airport locations. Regarding to economic aspects decision makers are forced 
to compare the alternatives based on the number of passengers that are going 
to choose the airport when doing a journey. 



Airport Choice & Competition - a Strategic Approach 



34 



As travellers are making their choice not only between different airports 
but also are able to take a land-based transport mode the resulting demand 
figures could not be evaluated in the unimodal context of the air service 
system. Nevertheless the total passenger figures are essential indicators for 
economic evaluation. 

Figure 13 shows the number of passengers forecasted for the year 2010 in 
seven scenarios. The highest number of passengers can be expected, when a 
system of two airports will be operated: One close to the city, serving 
national short haul flights and routes to some important European capitals. 

The other is situated up to 60 km from Berlin's city centre. It is more 
assigned for long, especially intercontinental hauls which covers also pure 
charter flights and direct flights to destinations, where the demand, 
originated at Berlin, has to be fed by national commuter-flights to provide 
satisfactory load factors (hubbing). At this stage we did not analyse the hub 
potential in detail as shown in the last subsection because here it is more 
important to ensure a 24 hour service and to avoid capacity restrictions at the 
airport itself as well as in the air corridors. 



3 JiSerbog West 
5 SchK efeld 

4 J^erbog Ost 

3 JiSerbog West 

2 Borlcheide 

1 Michelsdorf 

SSchSefeldSS! 

5 SchK efeld 
Tegel (city) 



^:i=f%^/^J:£S^im^ 



rA-^1tity;a-:^;iay/;;^^>ijJSg^gg|ii.» 



'kiT' >.'&^^4&'f?>^'^^•C^fjii'3ii^^^5^ U.S1 



E 
® 



Mill. 5 Mill. 10 Mill. 15 Mill. 20 Mill. 25 Mill. 30 Mill, 
in 1993 estimated amount of passengers for the year 2010 

Fig. 13. Passengers Berlin 2010 depending on airport's location 

From a macroeconomic point of view the modal split is also an important 
measure. Infrastructure investments in Germany must be evaluated accor- 
ding to well-defmed evaluation schemes. Herein the investor must apply cost 
benefit analyses beneath others. These processes require very detailed 
figures to assess a set of related impacts. Figure 14 depicts as an example the 
resulting mode choice pattern on the relation Beriin-Munich in the year 2010 
for the set of potential locations. 



Airport Choice & Competition - a Strategic Approach 



35 



3 jiBerbog \fVest 
5 SdiS efeld 

4 jSerbog Ost 

3 JiBerbog West 

2 Borkheide 

1 Michelsdorf 

5SchSefelds:S 

5S(±i5efeld 
Tegel (dty) 




g«agi»!^t4iadaBa^^ 



I waaaaatSBitegBBaiMSi^^ 



\t)t»iitssats^3mif^s»f>^»m!!\ fnn M mimtJUM 



)'dai''«'*iMwtoi»(id>fes^^ 



:aHi«WBi<»!WH»w>aca« » »f - ' >« M<iifl|^ 



0% ZQo/o 40% 

in 1993 estimated modal split for the year 2010 



■ air 
Drail 
Sroad 



y 



© 



60% 



Fig. 14. Modal split between Berlin and Munich 2010 
depending on the location of an airport 

Highest mode choice for air transport is given when there is an airport 
near Berlin's City centre, hke Tegel or Schoenefeld. Market shares up to 
almost 40% can be expexted then. On the other hand, when Berlin's new 
airport is situated about 50 km, the share of travellers by plane between 
Berlin and Munich decreases to about 20%. Obviously, the time needed to 
access the airport plays a major role on such short haul route. 

Regarding the environmental point of view decision makers will be also 
interested in the impact on natural and cultural resources. Measures for this 
field of interest could be derived e.g. from the modal split figures concerning 
the access and egress modes. Especially in dense populated areas as well as 
ecological sensitive areas the share of passengers using public transport for 
their ways from and to the airport are usefril indicators. Pollution could be 
directly derived from the absolute demand figures when applying distance 
related emissions to it. 

So when comparing several locations for a new airport regarding the 
change of passengers mileage in comparison to the status quo situation is a 
need. Separated by the different modes for access and egress as well as for 
air fransport, conclusions concerning energy consumption, vehicles 
emissions and noise can be drawn. This allows to set up some basic data for 
an environmental assessment concerning the different possible locations for 
a planned airport. Here (figure 15), airport locations with a large distance to 
the city of Berlin cause an increase of passengers-mileage up to 0.64 billion 
passenger-kilometres, although passenger-mileage of air transport 



Airport Choice & Competition - a Strategic Approach 



36 



diminishes in this cases, as air transport will lose some passenger on short- 
haul flights. On the other hand a new 'single' airport quite close to the city 
will cause a slight decrease of passengers mileage when sum up all modes. 
In this case passengers mileage of air transport rises up, while earthbound 
feeder transport to the airport goes down. Airport systems with one airport 
close to the city and a larger airport far off Berlin, have no remarkable 
influence of passengers mileage. 

Such an environmental analyses can be extended by computing the 
aircraft movements according to starting/landing routes divided into aircraft 
categories ( ICAO chapters). This allows to generate the noise distribution. 



Schoenefeid 

Micheisdorf 

BorkheJde 

Jueterborg West 

Jueterborg East 

Jueterborg West + 
Schoenefeid 

Micheisdorf > 
Schoenefeid 

Micheisdorf + 
Schoenefeid + Tegel 

Jueterborg West solo r 




1.24 



-3 Bill. Pkm -2 Bill. Pkm -1 Bill. Pkm Bill. Pkm 1 Bill. Pkm 2 Bill. Pkm 3 Bill. Pkm 
Pkm = Passenger kilometres 

Fig. 15. Change of passengers mileage depending on an airports location 



5.2.2. Accessibility of an airport 

In line with the Berlin example in the previous subsection the 
access/egress diversions of the public and individual transport modes were 
analysed. 

Modal split and assignment to feeder links for the new planned airport 
play a significant role when comparing several possible locations for a new 
airport. 



Airport Choice & Competition - a Strategic Approach 



37 



So, in the tables below one can see the number of passengers expected at 
the year 2010 on rail and road links to a new airport for Berlin, depending on 
its location. Scope of such a study is, 

- acceptance of airport-express-trains by the passengers, when the airport is 
situated quite far from the city, 

- additional capacity requirements on existing road links, 

- necessity of additional roads to the airport, 

- necessity of linking the airport to the forthcoming high-speed railway 
network in Germany. 

Table 2. Access/egress choice road 



Trips in mill. 


from the direction 
of Berlin 


from 
West 


from 
East 


total 
each Airport 


Scenario 


Reference: Tegel 








3,70 


and Schoenefeld 








9,12 


Schoenefeid South 








10,93 


Michelsdorf 


8,39 


0,80 


0,08 


9,27 


Borkheide 


7,54 


0,52 


0.26 


8,32 


Jueterbog West 


7,27 


1,20 


0.26 


8,74 


Jueterbog East 


7,23 


1,14 


0,24 


8,60 


Jueterbog West 


2,87 


1,06 


0,23 


4,16 


and Schoenefeld 








8,33 


Table 3. Access/egress choice rail 


Trips in mill. 


To/from Berlin 
Airport-Express 


Other directions 
with IC/ICE 


Other directions 

with regional 

trains 


total 
each Airport 


Scenario 


Reference: Tegel 








1,90 


and Schoenefeld 








4,05 


Schoenefeld South 


6,69 


0,17 


0,10 


6,96 


Michelsdorf 


6,44 


0,20 


0,07 


6,71 


Borkheide 


6.86 




0,18 


7,04 


Jueterbog West 


6,86 


0,45 


0,18 


7,48 


Jueterbog East 


6,74 


0,42 


• 0,17. 


7,33 


Jueterbog West 


3,47 


0,44 


0,17 


4,08 


and Schoenefeld 








3.44 



The main results were (with some differences between the specific 
airport locations) the 

- high acceptance of airport-express-trains, increasing with the distance to 
the city, 

- passengers flows to / from Berlin strictly dominate the total passenger 
demand, 

- connection to high-speed trains is useful when the airport is located in the 
south of Berlin (towards the agglomeration of Leipzig / Halle, 



Airport Choice & Competition - a Siraic\;ic Approach 



38 



Of course to evaluate such results in detail requires the necessary service 
characteristics, e.g. the public service frequency, which would exceed the 
limits of this paper. 

Due to the linked models for access/egress, one can analyse the number 
of passengers arriving at the airport by public transport. Compared to the 
previous figure one can see that the airport-railway station (high-speed 
intercity connections) has great impact on the access mode share. 



■ « % to 71 H 

■ so %l»M1fc 

□ 30 %!■ W« 
Qzd « MM% 
U 0Hl«20% 




+ ll^N 



Fig. 16. Market share of rail Frankfurt Airport 1991 

In 1991 the airport of Frankfurt already had one railway station close to 
the terminal. So the airport is connected to the Intercity-Network of the 
Deutsche Bahn AG. Beside this, the airport is also served by suburb trains. 
So using rail to reach flights starting from Frankfurt is quite common, as you 
can see in figure 16. The market share of rail to Frankfurt Airport raises up 
to a maximum of 75%, according to the distance to the origin of the 
passengers trip. A second railway station allowing more high speed trains to 
serve the airport of Frankfurt has just come into service in May 1999. 



Airport Choice & Competition - a Strategic Approach 



39 



Another point of view in analysing railway traffic to an airport is to show 
the number of people using trains to Frankfurt airport by regions, as we did 
that in figure 17. Many of the travellers start their trip in the hinterland of 
Frankfurt — although, there are also a remarkable number of people using 
ICE-trains along the high speed link Hamburg - Hanover - Frankfurt — 
Karlsruhe - Basel to reach Frankfurt's Rhein-Main-Airport. Even from 
Berlin, which is more then 500 km far away, about 100 Tsd. people per year 
take a train to Frankfurt Airport. 



Ra^May p«a»«n9wi 

■ WO 000 U 1 000 000 

■ 290 000 ta 900 000 

B too 000 \m 290 000 
Q 90.000 to 100 000 
Q 29000 to 90000 
29 000 




4' INN 



Fig. 1 7. Passengers to Frankfurt Airport by rail 1991 



Train operators can also be interested in origin and destination of railway 
users to or from an airport. In such a case more detailed information is 
needed. As an example therefore, the assignment of railway traffic to and 
from the Leipzig Airport, when equipped with an own railway station at the 
planned high-speed-link between Erfurt and Leipzig, is shown in the figure 
18. Assigning railway traffic to an airport helps to answer questions like 



Airport Choice & Competition - a Strategic Approach 40 

- Which towns have to be connected to the airport by direct trains? 

- On which lines existing trains offer sufficient capacity for the additional 
traffic? 

- Where have additional trains to be put into action, due to the increasing 
number of travellers? 

- Which revenues can be expected from serving an airport by rail? 



Wolmirstedl Burg 



18 12 



Wemigeroda 



GMlten 



Magdeburgir 



axjstino linl<s 

new link Erfurt- Halla / Leipzig 

Busshuttia from the station of 

Schkauditz 



116 



RoJ%iu 



18. 



V SchSebeckl y 

\\_ Calbe I ^ Wittenberg 

Bemburg«r • •Dessau ^f 

25^0 2 

K«»ien¥ Mf B 

55 I 

Aschersleben Cl 221 

Blankanhaim 68v 6 1 



WegeleberP^ 
Nordhausen 

\ 

20 Dis 

•-33 

Sangertiausan 



fi *. 221 tf 10 

^ I /Lands- 



-42-»-62-Ar 
Eislaben Haila J22. 



Lands- 
berg 



[538 



MtSihausen 

\ Erfur^ ^ '"'^ « 
Bad Langensalza • ^. _ , 

T "GroJaorbetha 

13 80 60 

Eisenach.^O-l-56-i ^•"*2"'«'» 
Gotha _ /Neudietendorf 

54 



28 
Bitterfeld 



Oelitzsch 



.1 3 

Eilenburg 



Torgau 



Amstadt 
.12 



«/^ 



^05 9« 13qi|4575 1/^^^^^^ ^^^^ 

Jau.zs^12iri^^^_ 22 19 

60 45 139 \ 

t tzei. t°«"" 



Suhl 



Apolda 



11 37 

^ P Jena 
la / 

Rudolstadt 



IZell 
30 80 



.30_» Oschal 

V\Ajraen \ 
27 
■V \ Riesa 

Grimma > 

17, 

Priestewitz 



GroFeringen "i' 

" 48-* ,. . 

Naumburg Gera A Neukieritzsch 



■11-^-Boma 



69 

■ AKenburg 



\ 



^ 



Chemnitz 



14 

t 

Dresden 



/ 



/' 



59 



GSnit 



Saalfeld 



•—10-^ Glauchau 
46 
Werdau L-yi-^ Zwickau 

^ Neumark • 



Plauen Hartasgrffi ^^ 
Mehlthauar^l 1-«_1 5_»_1 5-«^Relchenbach 



Fig. 18. Passengers (in thousand per year) by train to/from Leipzig Airport, 

2015 



5.2.3. Scenarios 



The last subsection presents some scenarios where the first and the 
second example are dealing with a network change of air services and its 



Airport Choice & Competition - a Strategic Approach 41 

impact on aiqDorts. While the first one is very simple and restricted to the 
inauguration of a new route at Berlin with the intention of showing the 
effects on the major competing airport in this market area, the latter example 
is quite complex in the actions considered and will show the effects on both 
airports involved. The third and fourth example are focussing on consumers 
elasticities due to ceteris paribus fare variations. While the third one presents 
the elasticity curves for different markets and trip purposes the last example 
displays a pricing strategy based on the elasticities stated in the third 
example. 

5.2.4. Route inauguration 

The German airport that serves the North American market the most is 
without doubt the Frankfurt/Main airport. This airport hosts the homebase of 
the former national carrier 'Deutsche Lufthansa AG' and is 'the hub' in 
Germany. As our scope is on consumers' behaviour in a competitive 
environment we want to study the impacts of changes on the supply side. 
Therefore the first step is to get an impression of the competitive situation of 
this major airport, which is displayed in form of Frankfurt's catchment area 
by Figure 19. 

As Frankfurt is offering a large number of services to North America and 
the earthbound access/egress possibilities are above average, its market 
dominance covers a wide area of Germany. The white spots at Frankfurt's 
hinterland do not indicate that Frankfurt doesn't play a role in consumer 
decision they just reflect on the one hand the good air feeder system to 
Frankfurt, which is used as access alternative instead of the landbased 
modes, and on the other hand the strong influence of competitors - e.g. non- 
stop service at the airport or other routes via competing hubs. 

Considering the status quo air network, an additional non-stop air service 
from Berlin to North America is installed in this scenario. The point of 
interest we want to show is the consumers' reaction to a new competing 
alternative which enriches the existing set of possibilities. Here the question 
about the demand elasticity plays a key role when arguments between 
airlines and airports are exchanged whether the originating market is big 
enough to install such a new service or not. Of course the transfer passengers 
will also partially use the new service, but for an airport manager it is more 
interesting to attract new customers than to shift air passengers from one 
flight to another. Obviously airlines will take another point of view. May be 
they compete with another airline and want to increase their market share or 
they want to enrich their service by another 0-D pair without losing the 
economic surplus at the already existing service of this market area. To 
analyse such effects we refer to the already discussed alternative or hub 



Airport Choice & Coinpetiiinn - a Strategic Approach 



42 



analyses in the second subsection. Anyhow for both groups it is important to 
know how the travellers react to service changes. 

Although Frankfurt Airport and the Berlin airport system (TXL, THF, 
SXF) are already quite far from each other, the catchment areas for 
origmatmg/destinating passengers are overlapping with up to 10% Berlin 
travellers using earthbound systems to access Frankfurt. The majority of 
travellers (90%) are choosing transfer services offered at Berlin via airports 
like FRA, AMS, LHR and CDG. Now the question is whether this situation 
can be mfluenced by an airport located at the border lines of Frankfurt's 
catchment area. 



■ more than 90% 


■ 80% to 90% 


B 60% to 80% 


Q 40% to 60% 


D 20% to 40% 


D 10% to 20% 


D 1%to10% 


D below 1 % 




p ZRK 



P INN 



Fig. 19. Market shares of Frankfurt Airport: destination North America 



Airport Choice & Competition - a Strategic Approach 



43 



^ 


■ -20% to -15% 
3-15% to -10% 

n-io%to -5% 

D -5% to -1% 
n below -1% 


1 p*;!^ 


F< 




Jl 




<A"S 




&M 


An TrCr"'^ppAD 


KSF-^ 


^^ 


)^'"'\>ij?^^ 


f^^ 




V^ 


^ 


''^ferrt-w 


5x^ 


Ay?p WXl^-FRAl 


fe 


$i 


"jj Tp^,9^^ 





F/g. 20. Market shares changes of Frankfurt Airport by regions if non-stop 
flights to North America are offered at the Berlin airport system 

In figure 20 we simulated the consumer reaction, when new non-stop 
flights to North America are offered at the Berlin airport system out of the 
point of view of Frankfurt Airport. As indicated in figure 20, Frankfurt will 
lose market shares in some areas belonging to Berlin's sphere of influence. 
The maximum decrease doesn't take place in Berlin directly — although the 
decline is up to 10% points which nearly diminishes Frankfurt's market 
share to zero — but in two counties situated in the south-west of Berlin 
where losses reach up to 20%. The reason for this strong reaction can be 
found in analysing the consumers' alternatives for reaching a destination in 
North America with and without non-stop flights offered at Berlin. 
Comparing the alternatives, consumers are making their choice in respect to 
their e.g. price and time elasticity which now results in passenger shifts to 
the new service at Berlin withdrawing them from Frankfurt. 

This reflects the obvious rule that the closer the starting-point of a trip is 
to an airport, the more people prefer this airport even, if then a transfer on 
their trip is needed, especially if the next airport offering non-stop flights to 
their final destination is far away. The main area of competition is at the 
regions where no airports are located. 

So travellers not originating in the vicinity of an airport have to compare 
very carefully their impedances to airports which offer non-stop flights to 
their final destination and the ones who do not. If e.g. the difference in travel 
time is less than or equal to the time it affords to change a plane, they will 
choose the new opportunity. So in the case of additional destinations for 
non-stop flights being introduced to the market, people react more 
sensitively to a new alternative by changing their starting airport. Figure 19 



Airport Choice & Competition - a Strategic Approach 44 

points out that Frankfurt's market share was up to 40% in such regions as the 
south-west of Berlin, although there were other airports closer situated but 
without a non-stop service to North America. The introduction of a new 
service at Berlin reduced Frankfurt's market shares by half as shown in 
figure 20. 

It has to be stated that the results differ by trip purpose and final 
destination so that the results shown are aggregated. In addition one has to 
be aware of the underlying access/egress infrastructure which is also 
mirrored by the catchment area of the airports. 

5.3. Secondary hub 

After the quite simple example a), a bundle of supply changes take place 
m the strategic scenario b). We assume that in addition to Frankfurt Airport a 
secondary hub in Germany at Munich airport will be established. The 
scenano consists of the following changes to the situation of the year 1991: 

1 . New additional intercontinental destinations are offered at Munich. 

2. The feeder network is extended to strengthen Munich's hub potential 

3. Some secondary destinations, offered in Frankfurt, are cancelled due to 
capacity constraints in favour of more flights to destinations with higher 
demand. 

Such supply changes create a new competitive situation between the 
airports Frankfurt and Munich where also other intemational airports are 
affected. In the following, we will focus just on the two airports Munich and 
Frankfurt. At first the changes in passenger volume at the two airports 
should be mentioned. While Frankfurt is losing 0.7 Mio., Munich gains 1.6 
Mio. passengers m total and on intercontinental routes Frankfurt loses 0.15 
Mio. which neariy can be attracted completely by Munich. Now it* is 
interesting to know how these passenger shifts can be explained e.g. what 
was the consumers behaviour? 

How consumers react in respect to the new situation can be summarised 
by the following five possibilities: 

1 . Travellers who used to depart from Frankfurt, now take off at Munich. 

2. Travellers who came to Frankfurt by earthbound transport to use a non- 
stop or via service, now take a feeder flight to Munich and reach their 
destination after a transfer. 

3. Some travellers who took a feeder flight to Frankfrirt, now use a feeder 
flight to Munich. 

4. Some other travellers who took a feeder flight to Frankfrirt, now go to 
Munich by earthbound transport to use a non-stop or via service. 

5. Travellers who used earthbound transport to reach Munich airport, now 
take a feeder flight to Munich. 



Airport Choice & Competilion - a Strategic Approach 



45 



It IS important to note that there is no general consumer reaction due to 
the complex structure of the bundle of strategic supply changes which in 
addition causes synergetic effects. As the five possibilities show the 
behaviour is always oriented to the individual situation reflecting a specific 
point of the elasticity curve. 

Hie change of passengers' demand on the flights between Frankfurt and 
Munich is also based on their consumer reaction described by possibiUty 2 
Information about the other kind of consumer reactions is given in figures 21 
and 22 which show the changes of the catchment area of Frankfurt and 
Munich caused by the new destinations offered at Munich for the market 
segments Asia and Africa. 

(bll] 
P 




Fig. 21. Market share changes at Frankfurt Airport: 
destination Asia and Africa 



Airport Choice & Competition - a Strategic Approach 45 

For Frankfurt Airport (Figure 21) a decrease of market share up to 4% 

l^^lV T ' """^^'^ °^ ^'^^^ ^°"^^ ^'^^"g^d in a wider circle 

around the airport location, but no significant change could be measured in 
he vicmity of the airport. The highest losses appear in regions situated close 
to other airports which are now connected to Munich by feeder flights (i e 
Saarbruecken (SCN), Nuremberg (NUE), Stuttgart (STR)) 

For the Munich Airport (Figure 22) a decrease of market share can be 
seen tor the areas around the airports of Nuremberg, Hof (HOO) and 
Fnednchsha en (FDH) which seems to be inconsistent with the extended 
schedules offered at Munich. The reason for this effect is listed as 
consumers reaction possibility no. 5 where consumers change from 
earthbound access to feeder flights. Additional market shares for Munich are 

tTT"^^^- "' '^" '''' °^ ^^"^^'^ ^"^ '^' ^^ffi'^ ^°nes at the border 
Iff f r^. uT^'" '' """'"^ ^y '^' "^^ intercontinental destinations 
offered at Munich that compete ^vith supplies at other airports. 




Fig. 22. Market share changes at Frankfurt Airport: destination Asia and 

Africa 

.v.^1 'f°'' '"'^^''' ■"'"'' '^'^' " "^^^ °" ^^^ possibilities which are 
available if one goes more into detail do^vn to the assigned air services by 

the al^n^r ^"'^'''''- ^T '^' ^"'"^ "'^ "°^ ^° ^^°^ '°^^^^ ^"d gains at 
the airports, the aim was to show the variety of the consumers' behaviour in 

respect to supply changes and that new competitive situations arise by 

complex strategic scenanos which include synergetic effects which can even 

be measured ex ante. The demand elasticities in respect to any air service 

vanable considered in the model specification (e.g. time, fare' fr queTcy 



Airport Choice & Competition - a Strategic Approach 47 

service attributes) allows to simulate and optimise strategies as well as to 

VIA (Last et al 1997) even (he role of earthbound feeder systems can be 
considered which as we saw in the previous sections cannot be n gkcted 
Fmally ,t has to be stated again that the results displayed, are agteLed 

5,3,1. Consumer's elasticity 

chaLVofZ'r'J fte practical relevance of the consumers elasticities the 
change of the market shares and travel demand of an alternative is computed 
m condtfon to an absolute change of the travel costs of tMs aZa" ve 
whereby all other conditions remain ceteris paribus equal As exal™ t^e 

f rTw^HTf*' f '"""t^* ^"^°" '" *^ ^"^ •'" - used andTphytd 
for three differem market segments, namely domestic, European »d 

ntercont,nen,am,e computed probabilities of the models will dfer for 

based on the specific ongin-destination (0-D) results 

The following figure 23 displays the change of passengers demand <v 
axis) based on the year 1991 when tariffs change (xlis). Ue z^^z^o c^ 
ordinates display the status guo at Hamburg in 1991. hca^cleaX brseen 

Lrif^^r '1'"^ °'^"''^ "^^^'"^^^ '^ *^ ^-^^l'^" because changesome 
tanffs have the smallest effect on their behaviour. Although increasing the 
tanff of domestic flights by DM 100 (~ USS 55) only 28% will^Wp ,he 
alternatives offered at Hamburg. At the same price chang * shafe of 
holiday makers will decrease by 37% and the pnvate travellers by f^! ^ 
he doryestic rnarket segment the air services have ,0 face a ston^ 
competition to the land based modes and in addition there are a lot of oth« 
mr alternatives around so that easily instead of Hamburg the a r^ort" 
Hanover or Bremen can be used iirpons 

give^ouTv'^ht' ' ^^ '°° *'"«' °^*= '^"ff f°^ European destinations 
give roughly the same picture just the losses of market shares are smaller 
(business 7%, vacation 25%, pnvate 38%). Surprisingly the resutafo 

S72™;) ft"^''ri"\*''"="'- *^"= *= "^^'^y -^"s - 

strongly (24 /„) followed by the business travellers (19%) the private 
travellers show the smallest effect (12%). 



Airport Choice & Competition - a Strategic Approach 



48 
The strong reaction of private travellers ha? tn k« . 



domestic 



european 
destination 



intercontinental 




variation of price per passenger 



♦"* ♦lOOOH «2D0OH 



Fig. 23. Share of travel de,na„d depending on cost changes 

paid by .h= company ort" ^ted X '"' *" *= ''*=' '= "^-"^ 

of the ticket price ^oZZyTu^tZ^^r^r' *"? *' *»«= 
trip, so that this is of minor rdevance AnThl 1 '°'^' '"P^^^.tures of the 
int^continental market is eL^:;S^:"aTow^^S:Te *■ '"^' ^°"^^'''*°" °" *' 

found'tto h"Uy Wpf .ore'L' 'Tf ^'"'^ "" "^^^^ ('^'«) " -- 

..m. period olS^i by^Ha^Jbtrn"^^^^^^^^^^ "=• » *= -- 

1.000 (~ US$ 555) ,0 Z,r^ff f!'^^"''^ "?""•'>' "P to DM 

differed w^ .^f.^^ o'SL i^^I^ tslea^dTnaf ^T' ^' "^T 
Joun,eys started from F^kftrt or Dus«ldorfatort ' '"^ 

should be madlslfcCTs of r.. ' I " °'"°"'- °"= ^''"^'^ 

.u. ioo.„, at the ^sJ:^::^:^^,^:^^^^^^r .. 



49 



Airport Choice & Competition - a Strategic Approach 

losiVm t^I f ^? '"•"'^"'^ '^^' '^' ^''''' °" '^' P^^^^nger side are not 
osses n total for the air market which will be shown later on in the nex 
subsection in more detail (see passenger shifts). A lot Tf travellers ^us 
choose another air alternative - mtramodality - only on short drsIefliiS 

subL'e blotr . °' T'' '''' ^"^"^^^^ destinations ^will be 
suDstituted by other ones due to the generation-distrib 



domestic 




european 
destination 



)ution approach used. 



intercontinental 



LSDOTh. . 





variation of price per passenger 



Fig. 24. Travel demand depending on cost changes 
Generally it has to be stated that changes of tariffs higher than DM 150 

ZZZ^''^X!':^'T'' ^"k' ''''-''''' ''' Precisiot^ft: result i 
aStth?^.?. " ? ^' "P^"'^^ ^^^" "^- observations are 

ava^able which catch consumer behaviour due to such large changes 

tnp between Hamburg and Frankfurt should be referred to because S 

the sense that nor large changes of air tariffs are unusual neither the 
elasticities found by the models can be neglected 

yJl%7'l Q7^'''''h °f^hetanff of about DM 50 (~ US$ 28) within a 
year (9.96 to 8.97) or a drop of the price by 30% and more as soon as 
another airline offers their service (e.g. Frankfurt - Beriin Hamburg 
Mumch or Munich - Ruhr area in 1997) or the anti trus Office cbfms 

m detail. In the light of the ongoing liberalisation, deregulation and 

qTt oTisX^ '': ''-'■ '''''' p"^^"^ ^^^^^^- ^^^^ th 

question is. How can aiiports participate at market procedures like airiine. 
already do for some time? Due to the enlarging capacity c^^ainf the 

nave to undertake the airimes and respectively the consumers have to face 
higher charges and / or fees. It will be a matter of time that airports wilf be 



Airport Choice & Competition - a Strategic Approach 50 

forced to turn to a more market oriented pricing strategy like peak & off 
peak pricing or the more general approach of slot trading to handle the spare 
capacity resources more efficient or say on the level of a real market price. 

Concerning the elasticities it is referred to the example of the Frankfurt - 
Hamburg 0-D passenger market which decreased by 4.8% from 1996 to 
1997 stated by airport statistics Hamburg. Within this time period not only 
the air tariff increased also the service frequency on this 0-D was reduced. 
Taking into account the average growth rate of about 7.4% in 1997 on the 
domestic German market the imaginable losses on this leg were 12.2% for 
the 0-D traffic. Obviously the growth rates were induced by the additional 
competition of airlines on several markets in 1997 where the prices dropped 
significantly. Therefore considering all effects the elasticities of the models 
tend to be conservative. In this context the question rises: Who can benefit of 
consumer elasticity by applying a market oriented pricing strategy, 
exclusively airiines? More and more the airports view the travellers directly 
as clients and apply aggressive marketing strategies to increase their 
attractivity (free or cheap parking and overnight stay, shops, play grounds, 
restaurants, high-speed rail access, etc.) to enlarge their catchment area. A 
new pricing strategy for the aviation side would be a natural enrichment of 
the existing marketing tools. In addition one could use a market oriented 
pricing as instrument to impose a price structure to meet political constraints 
like environmental benchmarks. 

By the way by increasing tariffs on an 0-D the total demand on this leg 
need not necessarily decrease if one takes into account the transfer 
passengers, who usually pay different prices. In the case of the Hamburg - 
Frankfurt leg which is dominated by the origin-destination passenger market 
the losses on total demand were 1.8% to the demand in 1996 because the 
share of transfer passengers increased by ca. 23%. 

Therefore to compute the effects on the leg level all itineraries on the 
total network have to be considered. Of course for each origin-destination 
pair as well as for each alternative serving an O&D such elasticity curvatures 
can be computed. Obviously it is wrong to concentrate on one airport and 
single services without considering the synergetic effects of a network and 
the competitive situation around. To face such and other complex problems 
like the air network or hub optimisation the airport choice models have been 
embedded in the system approach. 

5.3.2. Local pricing strategy 

The last strategic scenario deals with an increase of passenger fares at an 
international airport which might happen in order to meet environmental 
benchmarks (e.g. noise, pollution) or to manage scarce resources (e.g. 



Airport Choice & Competition - a Strategic Approach 



51 



parking positions, aircraft movements) efficiently. Again such an action will 
change the competitive situation between airports and the question arises 
how consumers react to the supply changes. This analyses is based on the 
elasticities shown in the previous subsection. 

Here we assume that the international airport Hamburg charges an 
additional supplement — to airlines or passengers — so that the travel 
expenses increase by DM 50 per passenger for any flight. The resulting 
question will be which kind of effects can be expected? Or the other way 
around, if one wants to reach a certain aim / benchmark which amount of 
money should be demanded from whom? In both cases the focus is on the 
price elasticity of demand. 

Figure 25 depicts the simulated market share losses for the airport 
Hamburg. The pattern results from passenger shifts to competing airports as 
well as travellers using earthbound modes (rail and road) as substitutes. The 
highest reductions of Hamburg's market shares can be found inside the 
extended area of Hamburg and in regions from where another airport (e.g. 
Hanover) is reachable in similar conditions, like the airport of Hamburg. 



r 




D 


0%lOl'/. 


U 


i%ioiy. 


U 


1%lo2% 


bJ 


2% to 3% 


■ 


3% to 5% 


■ 


5% to 7% 




P-^ 



f-. 




•E 
Z 
© 



Fig. 25. Market share losses of Hamburg Airport 1991: all destinations 

As we want to measure the competition we should have a look at the 
competitors. Where do consumers go to, which are the alternatives 
considered as substitutes, who are the winners or losers of such a scenario? 
Figure 26 summarises the passenger shifts between the competing 
alternatives. 



Airport Choice & Competition - a Strategic Approach 



52 



scena 



no: +DM 50,-/passenger on Hamburg Airport 1991 



HAM: -835Tsd 



to other modes: + 638 Tsd 




-200 



-150 -100 -50 50 

shifts of passengers in Tsd. 



100 



Fig. 26. Passengers shifts 

As the tariff increase is relatively high for short-haul flights the major 
effect is a shift to earthbound modes for domestic destinations. Here air 
services are competing with high-speed train services which serve a city-city 
pair nearly as fast as airplanes. 

Those airports connected with Hamburg by short-haul flights, like 
Frankfurt (FRA) and Dusseldorf (DUS) must be characterised as losers. But 
the total number of passengers on these airports decreases less than on the O- 
D flights because some of the travellers still reach these airports by plane 
just using a competitive airport like Hanover (HAJ) or Bremen (BRE). 
Others replace their former connecting flight (e.g. Hamburg - Frankfurt by 
car or rail trips to Frankfurt) and subsequently enlarge the catchment area'of 
these airports. 



Airport Choice cS: Competition - a Strategic Approach 53 

Aiiports situated closer to Hamburg may be considered as winners in that 
situation, ,f they are not connected to Hamburg by plane and, in addition 
provide a comparable number of destinations. Here, HAJ and BRE win more 
than 50 Tsd. passengers each, while at Kiel (KEL, in the north of Hamburg) 
there is only a little increase in the amount of passengers, due to the very fevJ 
destinations offered there. A special kind of winner, although the number of 
changing passengers is quite low, is Copenhagen Airport (CPH). Despite 
losmg passengers on the flights to and from Hamburg, the total number of 
people m Copenhagen increases. This result is caused by a combination of 
the two effects stated for Frankfurt and Hanover. 

Of course one can go even more into detail by analysing the consumer 
structure at the Hamburg Airport and how the segments are affected by such 
an mcrease of fares. Obviously business travellers are less price sensitive 
than holiday makers. But having a look at the passenger figures 
differentiated by tnp purpose, figure 27 indicates that although the number 
of travellers diminishes by 835 Tsd. (8%) the reaction of holiday makers is 
quite low - the total amount of travel expenditures is already quite high so 
that the extra charge does not have a tremendous influence on their decision 
- while tnps belonging to the trip purpose private (non-business trips up to a 
total duration of four days) are affected strongly. Due to the high time 
elasticity of business travellers this consumer segment is not affected very 
much. -^ 



6'507 Tsd. total | 



riva 
130/ 






Scenario: +dm 50,-/passenger: 
minus 835 Tsd passengers In total 



140/0 



private 
business - \ 



vacation 





^1 to other modes 



@ 



_^^ to other airports 



Fig. 27. Passengers at Hamburg Airport by trip purpose 1 99 1 



Airport Choice & Competition - a Strategic Approach 



54 
Again it can be stated that the elasticity of demand is dependent on the 

Now further analyses can follow concerning the effects on different 
routes, the aircraft movements, the environment or fma rconc^^^^^^^^ 
economic impact of such an action concerning the 

dec^se's 'thTnumtrr'''?' '" "^°^' "°^ ^^^^ P^--^-'^ ^--nt 
oecreases, the number of aircraft movements diminishes too a«; ficn.r^ ■:>« 

sows. Here (DM 50 extra charge per passenger), we M;aTed he^Xen 

ces by type of aircraft. When regarding the reductions by percentage fte 

721 whi h cr*" ": ''""' '" *= ^'"^^ °' '''^° P™P (represented bTATO 
72), which come mto service on short hauls only. On long hauls, which are a 

domain of panes hke the Airbus A340, the reduction of aircraft m™^ 

caused by the lack of passengers, is almost of no account, as that DmTo 

extra charge makes intercontinental flights only slightly dearer 



n total 




E 
© 



0% 



-5«/o 



-10% -15% 

aircraft movements 



-20% 



Fig. 28. Change of aircraft movements Hamburg Airport 1991 (percentages) 
When simulating rising passenger charges, the changing economic 
sima tion of an airport is an important point of view. So figured diXsfte 
development of aviation revenues. TTie actual revenues in 1991 at Sort 
of Hamburg were DM ISO Mill. With an amount of 6.5 Mill passen^s 
aviation revenues were DM 27.61 per passenger. When rising up pas'enS 
charges cetens panbus, by DM 100 (~US$ 55), the aviation revenues ll 
sum up to more than DM 300 Mill., although the number of passeng- 
estimated at Hamburg decreases and there are less aircraft movemrnT 



Airport Choice & Competition - a Strategic Approach 



55 



400 Mill. 



300 Mill. 



200 Mill. 




100 Mill. 



+ 



actual aviation revenues 
Ham burg Airport 1991 
(DM 27.61/passenger) 



+ 20 +40 +60 +80 +100 

[D^] additional charge per passenger 

Fig. 29. Aviation revenues depending on passenger charges 



Airporl Choice & Competition - a Strategic Approach 



6. ACKNOWLEDGEMENTS 



56 



in the,r business processes. In addition ain)ort surveys were nroviZ K 
all mtemational airports of Gennany whichlere Sed b^ thrctta^ 
Aerospace Center (DLR). Some of the results displayed areTxIractedTor^ 
work financed by clients who agreed to the presen^tion T^^^ap r £e7f 
was presented at Ae Air International Conference on T^nsportSon 

(C^U). TTie final version of this paper benefited from the Lful coi«2! 
ctcSHayli^^^ '--' -' °"-^ ^='»- - -' ^ -^^^^^n 

7. REFERENCES 

IZTe?-'cT^^-^1 ""'TT '^'^'^ ^^'y^'^' London; 1985 

Society Series B 26 19^ ^"'""^ of Transfonnations; Journal of the Royal Statistical 

''Tre?dli':97r" ^-^ ^^'" ^^^^^' ^^"^-^ ■ ^ ^«^-^-' -lysis; North-Holland; 

°"Se^nte\fltutf o'hT.^^^^^^^^ °^^--' °--^ Models, with 

<InteractionsBenv en Hgh Speed Rai^^ "nJ''^ ^'^ '^^'■'=«= ^OST 318 

fonhcomi^g ■ "*° "" '" '™^'°' ^'"'^•'^i'y of Moatreal, 12-1997; 

Distribution Modek with Quasi-Dirtct FoVmar rODPr ?%!" i""'"?' °™'«i''"- 



Airport Choice & Competition - a Strategic Approach 51 

Mandel B., Gaudry M., Rothengatter W.: A disaggregate Box-Cox Logit mode choice model 

of intercity passenger travel in Germany; The Econometrics of Major Transport 

Infrastructures; editor Quinet E., Vickerman R., pp. 19-44; MacMillian Press, London, 

1997 
Mandel, B., Gaudry M. I. J., Rothengatter, W.: A Disaggregate Box-Cox Logit Mode Choice 

Model of Intercity Passenger Travel ii; Germany and its Implications for High Speed Rail 

Demand Forecast; The Annuals of Regional Science, pp. 99 - 120; Springer Verlag; 1997 
Mandel B., Gaudry M., Rothengatter W.: Linear or nonlinear utility functions in Logit 

models? The impact on German high-speed rail demand forecasts; Transportation 

Research Part B, Volume 28B, No.2, pp. 91-101; 1994 
Mandel, B.: Schnellverkehr und Modal-Split; Nomos Verlag, Baden-Baden, 1992 
Manski C. F., McFadden D.: Structural Analysis of discrete data with econometric 

applications; London; 1981 
Sen A., Smith T.E.: Gravity Models of Spatial Interaction Behaviour; Berlin; 1995 
Tran L., Gaudry M.: QDF a Quasi-direct format Used to Combine Total and Mode Choice 

Results to Obtain Modal Elasticities and Diversion Rates; Centre de recherche sur les 

transports, University of Montreal; CRT-982; 4/1994 



A Study on the Flight Service Network for Incheon 
International Airport to be a Successful Hub Airport 

in Northeast Asia 

Kwang Eui Yoo and Yeong-heok Lee' 



ABSTRACT 

Incheon International Airport(IIA) is planned to open in about 18 months. Korean government 
has an ambition to make DA a major hub airport in Northeast Asia. The most essential and 
required condition for an airport to be a successful hub airport in a certain region is to have 
more efficient flight service network than the other airports in the same region. IIA should 
compete with Japanese airports to be a major hub in Northeast Asia because Japanese 
government also has a plan to expand greatly the airport capacity in Tokyo area and Kansai 
airport in Osaka. It is necessary for both IIA and Korean national air carriers to compose 
efficient flight service network considering hub competition with Japanese major airports. As 
the liberalization of international air transport industry would give more marketing freedom to 
airlines, they would plan the flight service neUvork and flight schedule based on market analysis 
instead of governmental regulations. In the economically liberalized environment, it is very 
required to analyze air passengers' flight choice behaviour in order to induce other carriers and 
passengers through IIA's attractive flight service network. Disaggregate model is more 
appropriate than aggregate model to analyze consumers' behaviour. The information derived 
from disaggregate choice model of air passengers could be utilized in devising efficient flight 
network and schedule plan. Value of travel time or trade off ratio between flight frequency and 
travel time which could be estimated from discrete choice model could be utilized for 
scheduling an efficient flight plan for airlines and composing efficient flight service network for 
IIA. 



' Both are professors at the department of Air Transportation in Hankuk Aviation University, Korea 



I . Introduction 

The aviation demand in Asian Pacific region has recorded larger growth rate than in other 
regions during last decade. In accordance with the rapid expansion of aviation demand in 
Northeast Asia, the construction of Incheon International Airport(IIA) was planned to meet the 
growing air transport demand of Korea and to play a role as a major hub airport in Northeast 
Asia. They should compose efficient hub-spoke flight network centered at IIA to make it a hub 
airport in the area. IIA is required to compete with other big airports in the same region to 
become a successful hub. Especially it is inevitable to compete with Japanese aiiports in Tokyo 
and Osaka because Japanese government also has a plan to add greatly airport capacity in those 
big cities and Tokyo is already known as a hub in Northeast Asia. 

Through the expanding open-sky policy in international air transport industry led by USA, 
airlines are predicted to operate to meet market needs. The national barrier will become less 
important than consumers' preference in the market when an international airline or an airport 
plans a flight service network. Therefore, the study of air passengers' behaviour in the target 
market should be treated as an essential base for flight service planning. 

The objective of this research is to study the way how to analyze the air passengers' flight 
choice behaviour and apply the findings of the analysis to air carrier's(or airport's) planning of 
flight service network. The area to study is air transport market in Northeast Asia region. To be 
more concrete, we will focus on hubbing competition of IIA with major airports in Japan. As 
they compete by flight service network, effective flight service network should be constructed 
through the scientific analysis of air passengers' flight choice behaviour. In this study, we will 
suggest a method to apply for planning flight service network so that IIA could win a 
competition with Japanese airports utilizing air passengers' choice model. 

There are several previous researches to utilize passengers' flight choice models in air 
transport planning area. Kanafani and Ghobria utilized air passengers' route choice model for 
their research concerning hub pricing of airport[7]. Benchemam also utilized discrete choice 
model to study air passengers' airport choice behaviour in UK[4]. Alamdari and Black studied 
passengers' choice of airline with logit modeis[l]. 

Following this introduction, section 2 is to review the air transport market in Korea and Japan. 
Section 3 will discuss hubbing strategies in air transport industr>'. Section 4 will introduce the 
method of empirical research and section 5 will be dedicated to main discussion of this study 
and section 6 is the concluding remarks of the study. 



II . Air Transport Industry in Korea and Japan 

This section will introduce the shape of air transport industry in Korea and Japan. However, 
this study does not introduce detailed information because it is not very necessary for the 
purpose of this study The following sub-sections are to review it roughly only mentioning the 
basic information related to this research. 

2. 1 The Policy for Air Transport Industry 

The policy for domestic air transport in Korea has been somewhat led by government 
regulation. Now, there are two scheduled airlines operating as private corporation; Korean 
Airlines(KAL) and Asiana Airlines(AAR). Korean government which wants to introduce 
deregulation to all industry has changed regulatory form of air transport industry in order to 
make it greatly deregulated. It can be expected that domestic air transport is going to be 
operated without governmental regulation in near future. For international air transport, Korean 
government is seeking different policy case by case. As they accept the suggestion of "open 
sky" from USA, the international air transport beUveen US and Korea is operated in 
economically liberalized environment. Airlines in this market can decide air fare, service route 
and service frequency without government intervention. However, the bilateral air service 
agreements with other countries except USA are more restrictive. They usually regulate service 
route and frequency. 

Japanese policy for air transport industry is a little more restrictive than that of Korea. 
Japanese government would like to lead air transport industry to the direction where they intend 
to drive. For international air transport, Japanese government also takes more conservative 
attitude than Korea, since they feel Japanese airlines are not so competitive, caused mainly by 
high cost. They want to keep on regulating air fare even though the degree of regulation is 
going to be less severe. However, Japanese government is considering the expansion of the 
routes of multiple designation. In general, they also try to adopt themselves to new wave of 
international deregulation of the industry. 

2.2 Capacities of Major Airlines and Airports in the Market 

There are several big scheduled airlines in Korea and Japan. In the aspect of capacity, JAL 
ranked the first place beyond compare and ANA ranked the second place by a little more 
capacity than KAL which ranked the third place. AAR ranked the fourtli and JAS ranked the 
fifth (refer to table- 1). 



Table 1. Major Airlines' Capacity in Korea and Japan (1997) 



Rank 


Airline 


Aircraft owned 


RPM(millions) 


World rank 


1 


JAL(Japan) 


143 


43,357.4 


6 


2 


ANA(Japan) 


137 


26,629.4 


14 


3 


KAL(Korea) 


119 


20,991.9 


18 


4 


AAR(Korea) 


45 


8,026.5 


39 


5 


JAS(Japan) 


88 


6,950.9 


- 


source: ' 


Major Airlines Profi 


les". Aviation week & space technology, Jan 


.1998 



There are three major international airports in the market; New Tokyo International Airport in 
Narita Tokyo, Kansai International Airport in Osaka, and Kimpo International Airport in Seoul. 
New Tokyo International Airport has one runway and has a plan to add two more runways. 
Kansai International Airport is also operating one runway and has a plan to add two more 
runways. Kimpo International Airport has two runways. However, in January 2001, all of the 
international flights will move to new Incheon International Airport which will have one 
runway at the opening date and another one in six months. Eventually, Incheon International 
Airport will be operated with four runways when they finish final stage of construction. 

The air passenger demand in Japan is concentrated in Tokyo area and Osaka area. New Tokyo 
International Airport and Kansai International Airport handled a major portion of international 
air passengers in Japan. In Korea, Kimpo International Airport handles almost all of the 
international air passengers. Table-2 shows the international traffic demand on these three 
airports. 



Table 2. International Passenger Demand at Each Airport(1996) 



City 


Airport 


International Passengers 
(thousands) 


Tokyo 
Seoul 
Osaka 


New Tokyo Airport 
Kimpo Airport 
Kansai Airport 


23,372 

21,271 

8,578 



source: 1. "Aviation shown by number(0j[^-C *■ ■& §i^)", Japanese Civil Aviation Bureau, 
1997 
2. "Aviation Statistics", Korea Aviation Development Association, 1997 



III. Hubbing Strategy in Air Transport Industry 

3.1 Introduction - Justification of Hub-Spoke Network System 

With the deregulation of air transport industry, airlines have altered their route structure to 
utilize their resources more efficiently and the hub and spoke flight network is proved to be 
effective. Hubbing occurs when airlines concentrate flights at a few airports which they use as 
collection-distribution centers for their passengers. Through hubbing, an airline could increase 
the number of connecting cities and flight frequencies with limited resources, which can be 
explained by fig. 1. 




fig. 1(a) Flight Service Network 
with direct Connection 



B 




. fig.l(b) Flight Service Network 
utilizing Hub-and-Spoke system 



Fig. 1(a) is to serve five cities with complete connection, by direct service only. As shown at 
fig. 1(a), ten (5C2 = 10) routes are required to supply complete connection with direct service for 
these five cities. Fig.l(b) is utilizing hub-and-spoke system, and it can be seen that only four 
routes are required to connect five cities by way of the hub city "C". If the city "C" is a big city 
generating large traffic demand, the flight frequency between "C" and other cities could be 
greater than that of beUveen small cities. Therefore, the passengers, for example, who want to 
travel between "A" and "D", have to transfer at "C", and this will enforce more travel time to 
passenger while the passenger can enjoy convenience by more flight frequency with hub-and- 
spoke flight service network. The air fare for connection flight usually cheaper than that of 
direct flight because airlines can reduce unit cost through high load factor. In many cases, since 
the route between hub city and a certain spoke citv' is for the purpose of transportation between 
hub and that spoke city, it may be considered as an additional revenue for the airline that earned 



from the passengers who travel between one spoke city and another spoke city by way of hub 
city. This will result in low air fare for the passengers who use connection flights. Therefore, the 
consequences for the passengers using hub-and-spoke system are the benefits from trading off 
longer journey times for more frequent flights, if necessary, and, on certain routes, the benefits 
from using cheaper flights. 

Even though it is normally accepted that travelers consider flight frequency, travel time and 
fare in their decision making procedure of transport choice, it has been proved that high 
frequency is usually more attractive to passenger than short travel time. In a competitive 
market, frequency seems to be a key variable and the S-curve relationship between frequency 
and market share is often cited. 

Since hub-and-spoke network systems are utilized in major continents in the world, multiple 
hub system serving between continents has been also developed(refer to fig.2). 




Fig. 2 Multiple Hub System 



3.2 Two Kinds of Hub 



As Doganis and Dennis(1989) proposed, it is reasonable to classify hubbing function of 
airports by two main kinds of hub; hourglass hub and hinterland hub (refer to fig. 3)[5]. As 
shown by fig. 3(a), through an hourglass hub, flights operate from one region to points in the 
opposite direction. Through a hinterland hub, short haul flights feed connecting traffic to the 
longer trunk routes. An hourglass hub usually only caters for connections in two directions, 
outbound and return. However, a hinterland hub serves as a multi-directional distribution center 



for air travel to and from its surrounding catchment area. 



A 



\/ 



\ 



. / 



fig.3(a) Hourglass Hub fig.3(b) Hinterland Hub 

IV. Research Method 

4.1 Introduction 

It is essential to study air passengers' behaviour for the planning of flight service network in 
the greatly deregulated air transport market. Disaggregate model is more appropriate than 
aggregate model in analyzing consumers' behaviour. This section will try to find the method 
how to apply the information derived from disaggregate choice models to flight service network 
planning. 

By traditional economic assumption, commodities are finely divisible with a change in price 
having an effect on the quantity of the goods demanded. However, this assumption does not 
hold for some commodities, transport choice being one of them. For such commodities, a 
change in price may result in zero consumption or unaffected consumption. When commodities 
are not finely divisible, marginal adjustments of consumption are not feasible consequences. 
Thus the individual behaviour of consuming discrete commodities is better represented by an 
individual choice function. 

4.2 Theory of Individual Choice Behaviour 

It is assumed that the individual attempts to choose from the range of alternatives the option 
that maximizes overall utility, when the hypothesis of utility ma.\imization is used as the 
decision rule of discrete choice. Individual k will select alternative i among a set of J 
alternatives if 



U,i,>Ujk (i5tj,j=l,2,3, ,J) (1) 

However, in repeated choice experiments, individuals have been observed not to select the 
same alternatives in the same situation, and different decision makers have selected different 
alternatives in the same situation with the same alternatives. This led to the development of 
probabilistic choice theory which attempts to explain these behavioral inconsistencies^]. This 
behavioral inconsistencies could be explained by random utility theory. In this random utility 
approach, the observed inconsistencies in choice behaviour are considered to be a result of 
observational deficiencies on the part of the analyst. The individual is assumed to select the 
alternative with the highest utility. However, the utilities are not known to the observer with 
certainty, and hence treated as random variables. Manski (1973) identified four sources of the 
randomness of the utilities, 

- unobserved attributes 

- unobserved taste variations 

- measurement errors and imperfect information 

- instrumental(or proxy) variables 

With random utility approach, the utility of the ith alternative for the kth individual can be 
partitioned into two components, 

Uik = Vik + E,k (2) 

where V,k represents the observable component, which also can be 

expressed as the systematic component or representative utility. 
Eii, is the unobservable component or random component 

It is assumed that the systematic component is the part of utility contributed by attributes that 
can be observed by the analyst. For the random component, the sources of randomness are those 
stated in the above paragraph. Since, in consuming commodities, individuals attain utilities by 
consuming bundles of attributes which define level of service, a relationship between utility and 
level of service can be defined, so that the observable component of total utility in equation(2) 
can be expressed as follows if a linear-in-parameters is assumed. 



Vik = ao+ aiXi + ajXi + + a„X„ (3) 

where, Va, = systematic component of utility of option i 
for individual k 

ao, %a„ = coefficients 

Xi, ;X„ = attributes of option i 

The coefficients (ao, ai, , a„ ) are assumed to be the same for all members of the 

population in equation (3). If different socio-economic groups are believed to have entirely 
distinct coefficients, then it is possible to develop an entirely distinct model for each subgroup. 
This is termed as market segmentation. However, socio-economic characteristics are often 
included in the model using an appropriate specification. In such a case the utility function can 
be expressed as follows; 

Uik = U(Zi,SO (4) 

where, Z/ = a vector of attributes of alternative i 

Sk = a vector of socio-economic characteristics of individual k 

4.3 Choice Model of Random Utility 

This subsection will introduce the basic theory of the random utility model, as the random 
utility approach is more consistent with economic theory. By combining probabilistic choice 
theory and random utilty theory, the following equations are obtained; 

Pik = Prob. [Uik>Ujk i^tj, j = l,2, J] (5) 

Pik = Prob. [ Vik+Eik > Vjk+Ejk i^j,j=l, 2, J] (6) 

Pik = Prob. [ Eik-Ejk >Vjk-Vik iT^j, j=l, 2, J] (7) 

where, Pn, is the probability of selecting alternative ifor individual k facing a set of J 
alternatives. 

It is important to stress that Vik and Vjk are functions of service attributes and are assumed to 
be deterministic. The terms Ejk and Ejk may also be functions, but they are random from the 
observational perspective of the analyst. It is usually assumed that the means of the random 
variable E's are zero, and any non zero means of E's are 'absorbed" into the systematic 



component of the utility function, unless noted otherwise. 

One of the most difficult arguments of random utility theory is defining a reasonable 
functional form for V. Ben-Akiva and Lerman proposed two criteria for selecting functional 
form; (1) the function to reflect how the various attributes in the alternative set influence utility 
(2) the function that has convenient computational properties that make it easy to estimate their 
unknown coefficients[3]. In most case, functions of linear-in-parameters are chosen. As for the 
functional form for the distribution of random component E, different assumptions regarding 
the distribution of E, lead to different choice models being developed. Although several models 
for the multinomial choice situation have been developed, multinomial logit (MNL) is the most 
widely used multinomial choice model . 

4.4 Application of Stated Preference Techniques 

Often it is not easy to calibrate an efficient discrete choice model with revealed preference 
data because there is not sufficient variation of all variables of interest, and there are also often 
strong correlation between variables or between variables and other invisible factors. Stated 
Preference(SP) techniques which allow the researcher to experiment, can offer a solution to 
these problems. With clearly defined attributes and attribute levels, SP experiments can give 
researchers the chance to have sufficient variation of variables of interest, and an orthogonal 
design which ensures that the attributes presented to respondents are varied independently from 
one another, avoids multi-collinearity between attributes. 

4.4.1 Introduction to Stated Preference Techniques 

SP methods which were originally developed in marketing research in the early 1970s have 
been applied in the empirical analysis of transport-related choice behaviour since 1979. Though 
these techniques were severely discredited at their beginning, by the end of the 1980s, they were 
perceived by many researchers to offer a real chance to solve the problem related to transport 
demand modeling. 

Kroes and Sheldon (1988) described SP methods in transport research as a family of 
techniques which use individual respondents' statements about their preferences in a set of 
transport options to estimate utility functions[9]. The options are t>'pical descriptions of 
transport situations or contexts constructed by the researcher. Generally, SP techniques can be 



defined as all the approaches which use people's statements of how they would respond to 
hypothetical situations. 

4.4.2 Advantages of SP Techniques 

Transport planners need to know the likely effect of any planning strategy they consider. 
However, the traditional methods using revealed preference data cannot provide good quality 
information on travel demand and travel behaviour mainly because there is insufficient 
variation in the variables of interest to produce statistically significant models, and further, such 
variables are often strongly correlated. Moreover, revealed preference methods cannot be used 
to evaluate demand under conditions which do not yet exist. SP techniques, however, allow the 
researcher to experiment the consumer behaviour under various conditions, offering an effective 
solution to such problems. The advantages of SP techniques over revealed preference (RP) 
methods are summarized as follows[l 1]: 

(1) RP : Observations may not vary sufficiently for the construction of an accurate statistical 
model and the variables may also be correlated making it difficult to estimate model parameters 
reflecting the proper trade-off ratios. 

SP : SP techniques can ensure data of sufficient quality to construct a good statistical model 
because the researcher can control the choices offered to respondents. 

(2) RP : The observed behaviour may reflect factors which are not of interest to the policy 
maker. In addition, the effects of the variables that are of interest may be "swamped" by these 
other factors. This is a particular problem with "secondary" qualitative variables. 

SP : Due to the control available to the researcher, the effects of variables of interest can be 
isolated from the effects of other factors. 

(3) RP : There is no information on how people will respond in situations where a policy is 
completely new. 

SP : Where a policy is completely new, so that no RP data is available, stated preference 
techniques may represent the only practical basis for evaluation and 
forecasting. 

(4) RP : To obtain adequate observations of behaviour, very large and therefore very expensive 
surveys may have to be carried out. 

SP : Since each slated preference interview produces multiple observations per individual, 



efficient statistical models can be developed from much smaller sample sizes. 

V. A method on IIA's Strategic Flight Service Network Planning to Win Hub 
Competition in Northeast Asia 

This research reviewed air transport industry and introduced the hubbing strategies in the 
industry and discrete choice modeling. In this section, we will discuss how to utilize the 
information which could be derived from discrete choice model for IIA's strategic flight service 
network planning to make it successful hub airport in Northeast Asia. 

5.1 Information derived from Discrete Choice Model to be Utilized for Flight Service Network 
Planning 

This paper will research the method to utilize the information derived from the analysis of air 
passengers' flight choice behaviour for flight service network planning of IIA. Discrete choice 
model is useful to understand passengers' choice behaviour. Under the assumption that some 
utility functions concerning air passengers' flight choice have been calibrated, the methods to 
utilize the information derived from the models to IIA flight service network planning will be 
presented in this section. 

Through the previous studies in the industry, it has been identified that flight frequency, air 
fare, and travel time are the major attributes to air passengers' flight choice behaviour[12]. If a 
discrete choice model is calibrated using these three attributes and equation (3) of this study, the 
results may be presented as follows; 

V = a<, + a^ FARE + a, TIME + af FREQUENCY (8) 

Even though the magnitude of individual coefficient of equation (8) is important to estimate 
the weight of each variable considered in consummers' choice behaviour, this study would try to 
utilize relative importance of pair of variables, which can be estimated as the ratio of any two 
coefficients. The reasons to utilize relative importance of variables are as follows: (i) The 
passengers' flight choice or route choice is decided comparing each variable. That is to say, 
relative importance of variables becomes significant factor when he/she decides to choose an air 
trip an\"way. (ii) Especially, the model coefficients estimated from SP data are not proved 
appropriate to be utilized as absolute value. Instead, the SP model is useful for seeing the 
relative importance which can be estimated by comparing the absolute value of coefficients[ll]. 



There can be three ratios estimated by comparing any two variables with each other if a model 
is composed of three variables; air fare, travel time and flight frequency. The three ratios and 
their significance could be explained as follows, utilizing the quotation of the coefficients of 
equation (8): 

(i) RATIO- l;a/ac 

where; a, is the coefficient of travel time variable 
Oc is the coefficient of travel cost(airfare) variable 

(ii) RATIO-2; a/a^ 

where; aj is the coefficient of flight frequency variable 
Oc is the coefficient of travel cost(air fare) variable 

(iii) RATIO-3; aA 

where; a/ is the coefficient of flight frequency variable 
a, is the coefficient of travel time variable 

RATIO- 1 is the ratio between travel time value and travel cost value. This ratio is the most 
frequently utilized relative importance in transport studies, which is usually mentioned as value 
of travel time (VOT). The relative importance of flight frequency to air fare can be calculated 
by RATIO-2. RATIO-3 is the ratio between the coefficients of flight frequency variable and that 
of the journey time variable. This ratio is usually considered as a trade off betxveen service 
frequency and travel time and can be utilized when they consider the choice between direct 
route system and hub-and-spoke system. 

5.2 Methods to apply the Information derived to IIA's hubbing strategies 
5.2.1 Application to Hinterland Hub Strategies 

It is essemial factor to have plenty of short-haul flights in catchment area in order to be 
successful hinterland hub. In addition, they should try to reduce transfer time required to change 
aircraft for the connection beUveen short-haul and long-haul flights. IIA should try to increase 
the flight frequency considering the competition with Narita and Kansai Airports. Since the 
major airlines in Korea have much less capacit>' than Japanese major airlines to have enough 
flights to compete with, it is desirable for Korean airlines to utilize the alliance with Chinese 



airlines and Japanese regional airlines. With limited capacity, Korean airlines and IIA airport 
operator should try to supply more efficient flight service to make IIA a successful 
Northeastern Asia hub airport. In order to achieve such an object, this study suggests that the 
ratios of coefficients of discrete choice model can be utilized as follows. 

They can utilize RATIO-1 to set air fare and to decide aircraft type to introduce. It is basic to 
introduce cheap and slow aircraft for the routes which reveals low VOT (value of RATIO-1) 
and to introduce expensive and higher speed aircraft for the routes which reveals high value of 
VOT It is required to consider RATIO-2 in order to compromise the level of frequency and air 
fare. For the routes which have higher value of RATIO-2, they should try to increase flight 
frequency suff-ering low load factor. Low load factor may lead to high price inevitably if airlines 
seek to recover the operation cost. On the other hand, for the routes which have lower value of 
RATIO-2, it is eff-ective to reduce flight frequency, which could result in higher load factor and 
lower air fare. 

It is a normal practice that the routes which have large portion of business passengers would 
have higher value of RATIO-2 than the routes mainly composed of leisure passengers. If any 
routes are operated for mainly leisure passengers with small amount of demand, and if there is 
significant local traffic between cities near each other, then combining destinations on one or 
more spoke can be effective (refer to flg.4). In the case of which RATIO-1 is very small, this 



kind of routing strategy is desirable. 




-> 



fig.4 Hub and Combining Spokes Network 
( Cities "P" and "Q" are combined spoke in the figure) 
It would cost some expenses to improve operational standard to reduce connecting time on 
hub airport. The airline and airport operator should decide the level of cost to invest in order to 
reduce transfer time and they should set the level of air fare to recover the invested cost. It is 
useful to utilize RATIO-1 to optimize these t^vo variables; travel time and travel cost. However, 
for the passengers originating from the cities where direct connections to long-distance major 



cUies a. tapo.ib,e o. i„co„ve„ie„, i, „., ,e desirable ,„ i„,™d„ce low fare and high 

5.2.2 Application to Long-haul Flight Semee Network Planning „f ,ia 

We consider the long-haul flights as flights to se„e inter-continents routes and there a™ 
re.at.ve,, large dentand between Korea .d North Anterica^urope. However, the Eurpe!^ 
™.s are s.g„,«eantl, regulated b, bilateral air service agreements and the demand to eZ 
. far less than that of USA. For this reason, this stttdy would discuss the flight netwl 
Planning strategies on the routes to Nor«, America only as long-haul flights 

Because of the inferiori^ in the aspectof airline capacity as well as the magnitude of demand 

a11 flA an?K ^ '"""' "^ ""'" """"" '"'' ""^ ^'J" =■"- i" No'th 

sZ: g ; '"""" '""" """' '" ""- ™-"' "-- •» "'" -- «= 



SEL 



San Francisco 



N.Y. 




W.D.C. 



fig.5 Long-haul flights between Seoul and Major Cities 



in North America 



SEL 



San Francisco 



N.Y. 




fig.6 An Example of Revised Long-haul flights 

The new system is the one which impose hnbbing concept. The old one which has ditec. 
connection to many cities ,vith low frequency may be suitable for the leisure traveler and for 
cargo. However, such low frequency services do no, offer the flexibility required by business 
communny. The new system is ,o concentrate on high frequency services on the dense routes 
For the other cities, connections are provided either by change of gauge equipment or allied 
partner a.rline's own local service. However, no one can calculate that the new system is better 
than old one for the airlines' or air passengers' welfare. „ is necessaty to estimate RATIOs 
defined in this study, and apply i, for the decision making. 

The discrete choice models should be calibrated for individual route separately. ,f the RATIOS 
esttntated from discrete choice model of each route reveal that passengers of each route prefer 
evenly direct flight with scarce frequency of flight, which means high value of RATIO-l and 
low value of RATI0.2. the old system is more appropriate than new one. However if d,e 
RATIOS estimated from discrete choice model of each route are signiflcantly different or they 
show that high frequency with longer travel time preferred, then i, is justified to introduce new 
system. 

To introduce new system, they estimate RATIOs from discrete models of each route for 
example, route to Los Angeles, and route to San Francisco. They need to concentrate the flights 
on the route of the higher value of RATIO-l which is selected as trunk route. The reason why 
they should utilize RATIO-l is that RATIO-l is the most seriously damaged one by 
•nten^ediate stop. That is to say, the passengers who have higher VOT should be provided with 
direct service. 

In addition, Korean airlines could utilize codesharing or other alliance techniques with 
Amencan airlines to compose efficient flight net^vork. Especially, the connection flight between 
fore-gn hub and spoke cities in USA should be operated by some of US airlines which allied 
wuh Korean airlines. Therefore, an airline which has scheduling power on foreign hub airport 
should be pointed as alliance partner. The transfer time between Trans-Pacific long-haul flights 
and short-haul flights connecting to some cities in US should be considered utilizing RATIO-l 
Th.s IS because there is considerable competition with direct flights. 



5.2.3 Application to Hourglass Hubbing Strates 



les 



It .reasonable in the aspect of geographical position for IIA to take a role acting as an 
ho.g.ass hub to connect the air passengers traveling between Southeast Asian Cities an Cities 
.n the West Coast of USA. Actual., significant nu.ber of passengers traveling the cities o 
these reg:ons are transfe.ed at Ki.po Inte.ational Airport. This traff.c could be handled as the 
s.xth freedom a.r transport and low fare could be applied. Anyway, to set the air fare and flight 
frequency RATIOs should be utilized. To compete with direct flights between Southeast AsL, 
C.es a.d West Co.t Cities in USA. IIA should offer low fare and high frequency which can 

ffset the negative effect caused by longer travel time. RATIO- 1 would be effective is setting air 
fare and RATIO-3 would be effective in setting the level of flight frequency. 
The results found through the discussions of section 5.2 could be summarized like table-3. 

Table 3. Summary of IIA's Efficient services Network Strategies 



Major factors of competition 



Hinterland Hubbing 
Strategies 



Long-haul flight 

Service network 

Strategies 



Plenty of short-haul flights 
Minimum Connecting Time 
Introducing efficient aircraft 
type 



Hourglass Hubbing 
Strategies 



Integration of long-haul flights 
to concentrate on competitive 
routes 



Applied RATIOS 



RATIO-] : to set air fare and 
aircraft type 

RATIO-2 : to compromise the 
level of fi-equency and air fare 



6th freedom 



Increasing 
transport 

Low fare and high frequency 
service 



RATIO-I,RATIO-2: to select the 
routes which IIA concentrates on 



RATIO- 1 : to set airfare 
RATIO-3 : to set the level of 
service frequency 



VI. Concluding Remark 



W„h ... ,„„d toward. ,ib,ra,..io„ i„ air ,ra„spo„ indus,^, air passengers wi„ ,ave .ore 
op..o„s for ,he,r .ravel. ,„ a .ore flexible p,a„„i„g environmen,, air .ranspon system p,a„„ers 
airpor, operators and airline operators will need to know the consumers p„ferenee. „A which 
has an ambition to be a hub ai^„ in Northeast Asia should study the consumer, behaviour 
and ut .e the results for flight service network planning. Discrete choice models would be 
useful for analyzing air passengers' flight choice behaviour. Section 5 of this study introduced 
several ways to apply the information derived from air passengers for HA's hubbing strategy. 



REFERENCES 



[o trnT: V"' "^^'' ' "''''''''''' ''-''- ''^''- ""^- ^-P--''-: The Use of 
Logit Models", Trar^sport Review, vol. 12, No. 2, 1992 



1997 



(21 AWiST. -Major Ai,i„e. P„mes-, .w„„.„ P..*.„«^,„ ,,,,„„,„^_ ,„ ,3^ 

[3] B=„.Akiva, M. and Le™a„, S., D,.cre,e CMce Analysis. MT ?,,s.. USA, 1985 

[4J Benohe^a., M.^,i.,^,« ^,,„,„„„, ^„^^^, ^^^^^^ ^ 

Loughborough University, 1986 '^' 

ra Doganis, R. and Dennis. N., "Lessons in Hnbbing", AJrl,„. Business. IVIarch ,989 

C«J Japanese Civi, Aviation Burean, .w»„o„ s.o™ ,, ™„,„,^^, ,^ ^^^^ ^^^_ 

m Kanafani A. and GhobHa,. A., -AWine „„bbi„g.So™e taphoations For Airpor, 
Economics", Transpor, Resewxh A.yol 19A, 1985 



Seoul, 1997 



[8J Korea Avialion Developmen. Association, A,ia,ion S,a,is,ics(m£mm. 

m Kroes E. and Sheldon, R. "Stared Preference Methods, An Introduction". Jo.„., ,j 

im Manski, C, ne Analysis ofQua,Ua,i,e CHoic,. Ph.D Tresis, MlT Cambridge, USA, ,973 

(lij Peat^ain, D., Swanson, J., Kroes, E. and Bradley, M., S,a,e, P^f.r.„c. T..Hni,..s- A 
C..*.^.«,„,SteerDaviesOreeveandHagueConsu,tingGroup,Lo„don, 1,9, 



AIRFREIGHT FROM A PROCESS CONCEPT 



Prabal ACHARJEE 
Kenth LUMSDEN 

Department of Transportation and Logistics 
Chalmers University of Technology 
S-412 96 GOTEBORG, Sweden 
Fax: +46 31 772 1337 
Phone; 446 31 772 1326 
E-mail: prabal@mot.chalmers.se 

Phone: +46 31 772 1345 
E-mail: kri @mot.chalmers.se 



ABSTRACT 

Airfreight has gained a significant rise in the market as the need for fast and efficient 
transportation has increased over the years. Airfreight is transported either through 
pure freighters and trucks or through the belly of the passenger planes. The process 
of transporting freight through the belly is rigid through several factors like short 
tumaround time, priority issues, ramp congestion at peak hours, aircraft types etc. 
Belly's flexibility lies in the frequency of the flights to make it theoretically possible 
to deliver the goods on the same day. Pure freighters are not constrained by as many 
of the hindering factors as does the belly but not similarly flexible to manage 
deliveries on the same day. The strategy of the integrators is purely to deliver the 
goods 'overnight' and thus they are rigid in their services and processes. This paper 
analyzes the airport processes related to belly-airfreight and also the possibility of 
utilizing the belly more efficiently. The paper also investigates if a more efficient 
utilization of the belly at the daytime can generate a new concept of processing 
airfreight by achieving a more significant share of the market. The paper is 
empirically based on qualitative and quantitative data generated from the airport 
process operators. 



1 INTRODUCTION 

Airfreight is normally defined as freight with high-speed delivery. Goods originated from the 
shipper gets the fastest carriage (as airfreight) on its way to the final receiver. Need for fast and 
secured transportation has considerably increased under the last decades as a result of new 
layouts in industrial activities, for example, customer-order driven production and centralized 
warehousing (Lumsden, 1998). The airfreight market is increasing by 12-15% every year. In 
Sweden, for instance, export by air has been increased by 18-20% under 1997 (Transport och 
Hantering). In spite of the fast growth of the airfreight market, there is a lot of conservatism in 



the branch. Although the growth of airfreight has been more in compare to the growth of the 
passengers, airfreight is still considered as a by-product in the line-based traffic (Dahllof 1997). 
Earlier, airfreight was never forecasted to be an industry. A proof to this argument is the 
architecture of the airports in general, which is not very friendly when it concerns cargo 
handling. Faster and secured delivery is a prerequisite for the existence in the market for many 
actors. Often airfreight is the only realistic alternative. 

1.1 Airfreight actors 

Airfreight can be transported in different ways - via pure freighter, via belly of the passenger 
planes or via trucks. The customers do not need to know which of the three ways the gods are 
transported by. What they are interested to know is that the goods are delivered fast and on time 
as promised. 



Actors 



Type of operation - 





Airfreight actors 






























Airlines 




Integrators 




























® 

Separately or partly 
owned cargo 
organization 


® 

Separate cargo 
division within the 
same organization 


<3> 

No separate cargo 
division but cargo 
handling within the 
passenger division 






Only cargo 
organization 



Figure 1: Airfreight actors 

The airfreight actors can be divided into two kinds. The traditional actors i.e. the airlines and the 
latterly emerged integrators. The airlines have in principle three different strategies in their cargo 
operations. Cargo could be: 

1. a wholly (or partly) owned separate organization within the same brand name; or 

2. a separate division within the airline; or 

3. a not separated division in the airlines. 

The first type of cargo organization is generally independent or tends to be independent of the 
passenger unit for their services. They generally plan their operations in such a way so that they 
can utilize their own resources as much as possible. They buy specific services from the 
passenger unit. The second type of cargo organization do not own resources themselves and pays 
to the parent company price for utilizing its resources. Cargo activities in the third type do not 
constrain themselves in any organizational boundary. There are companies that do not have any 
separate division for cargo. They set price for the freight and whatever is shipped is considered 
as a contribution to the business. Airfreight is a by-product for these companies and considered 
as 'better than nothing'. 



The integrators, on the other hand, have integrated freight flow all over the chain. Their strategy 
is to deliver the goods overnight. They own all the assets (e.g. different vehicles, information 
system, etc.) all the way from the shipper to the final receiver. Consequently, they have a good 
control over the flow and can effectively (but in a rigid way) deliver the goods. Their quality 
service has made it possible to have a continual increment in their market share. Although their 
core business was to make door-to-door delivery of small packages, they are continuously 
expanding their operations with additional services of shipping bigger volume of freight. 

1.2 The importance of belly capacity 

The most economical way to transport the airfreight is through the empty belly of the regular 
passenger aircraft in the line-based traffic. This is obvious if seen from the transport provider's 
point of view. The cost is then minimum and the income is maximum. There is also an advantage 
of the belly utilization and that is - the flights have a very good frequency. The customer service 
(for freight) could be increased if the belly is used smartly. The problem although with the 
passenger planes is that these are designed to transport people, not freight. Here, a priority list is 
maintained where freight, unfortunately, is the last to enter an aircraft. Moreover, there are other 
factors that hinder a better belly utilization, e.g. — departure time, type of aircraft, uncertainty of 
the amount of passenger baggage, turnaround time maintenance, congestion at the ramp at peak 
hours etc. (Acharjee et al, 1999). In Sweden, for instance, the amount of goods accompanying a 
regular domestic flight is very small (<100kg/flight) (Larsson 1998). But for international flights 
the amount of accompanied goods is larger. 



2 AIM OF THE PAPER 

The objective of this paper is to find ways to promote the possibility to offer a shorter door-to- 
door delivery time through utilizing the empty belly capacity in passenger aircraft. The paper 
also investigates if a more efficient utilization of the belly at the daytime can generate a new 
'overday' concept of processing airfreight and thus achieve a more significant share of the 
market. 



3 METHOD 

The research approach, to fulfill the objective of this paper, is of a combination character. It is a 
combination of systems and analytical approach. For a process investigation (which is the first 
part of this study) possible relationship between systems must be understood. In the method of 
proceeding towards that goal the actors will be identified and their activities in the different 
systems (connected to each other in the whole process of line based air-traffic) will be clarified. 
The problem factors will be analyzed and the possible strength of those detected factors will be 
point out. In traditional analytical research the test of hypotheses is vital. Analytical approach in 
this case is not from a "hypotheses testing' perspective but of an explorative nature. The analytical 
approach here is the verification of the strength of the detected problems and also, measuring of 



possibilities to tackle the problems through inputs from more actors in the process. The steps are 
as follows: 

• Identify the actors in the airfreight flow and understand their activities in line-based traffic 

• Understand factors that hinder belly utilization 

• Analyze the influence of those factors on goods flow and estimate possible strength of those 
factors. 

• Verification of the estimated strength through questioning the airfreight actors on the level of 
control over the hindering factors. Transform the qualitative data to quantitative data to conclude 
how belly could be utilized better and if a better utilization might generate a new concept of 
processing airfreight - that is, utilization of the passenger aircraft belly at the daytime. 

4 PROCESS IDENTIFICATION 
4.1 Actors and activities 

A number of actors participate in the process of the airfreight flow in line based traffic. The flow 
is a combination of activities taken care of by a few or all of these actors — customers, 
forwarders, terminal handlers, ramp handlers, and air companies. In most of the cases the 
forwarders deliver the freight to the airport. The freight can even arrive at the airport directly 
from the customers, means that the customers take care of the transportation themselves. The 
terminal handlers receive the goods and prepare (palletize or pack) them to be taken by the 
ground handlers to the aircraft. The goods can even be trucked from the airport to shorter 
geographical territories. In that case the terminal handlers load the goods to the trucks. 

I 

I 
I 





,^'' 


^ . 


"■■4 


Lit 


le-base 


;d traffic 






Customers 


,--' 


Forwarders 


Terminal 
handlers 




Ground 
handlers 


Aircraft 






^ 

\ 













-^1 Trucks 



Figure 2: Actors in the process of airfreight flow 



4.2 Hindering factors 



In order to understand what causes a poor utilization of the belly, a study of the freight process 
were made at the Landvetter Airport in Goteborg (the second largest airport in Sweden). In the 
first phase, the study looked at the process of how the goods are treated on its way from the 
airport terminal to the aircraft. Through interviewing the personnel engaged in operations, factors 
that generate a low utihzation were noted. It was understood that a poor utilization of the belly is 
caused not only by certain operational inefficiencies in the airport but also by various other 
factors. In fact, the majority of these factors lie outside the operational inadequacies in the 
airport. In the second phase of the study a number of airline operators were interviewed. The 



airline interviews were extended later to the largest airport in Sweden (Arlanda Airport, 
Stockholm) to have a better view on the problem as a whole. 



Customers 



> 



Forwarders 



Factor Group ) 



Terminal 
handlers 



Ground 
handlers 



Factor Group 2 



Aircraft 



Factor Group 3 



Arrival of goods 
ai the airpon 



Departure of goods 
from the airport 



Figure 3: Different hindering factor groups and their position in the system 

The hindering factors found out in the whole study are divided into three main groups (figure 3). 
Each of the groups is further divided into different subgroups of factors (table 1), named as 
"Factor Group 1', "Factor Group 2' and "Factor Group 3'. A hindering factor, termed as Triority 
maintenance', has been placed in "Factor Group 1' under subgroup 'strategic factors' although this 
factor concerns the physical handling of goods at the airport. This factor could be argued to be 
placed in "Factor Group 2' under 'operational factors'. The reason why it is placed under 'strategic 
factors' is that 'Priority maintenance' is considered more as a strategic than as an operational 
issue. 





Factor Group 1 


Factor Group 2 


Factor Group 3 




Outside the airport 


At the airport 


Aircraft 


u 
b 

B 

r 
o 
u 

p 

it 


• Market-related factors 

• Strategic factors 

• Delivery-related factors 


• Operational factors 


• Unavailability issues 

• Incompatibility issues 



Table 1: Factor groups and their subgroups along with their position (origin) in the system 



4.2.1 Factor Group 1 

'Factor Group 1 ' denotes hindering factors that are not caused by any operational inadequacy in 
the airport. This can otherwise be expressed as factors not related to the physical handling of 
goods in the airport. This group contains three subgroups of factors. These are 'market-related', 
'strategic' and 'delivery-related' factors. 



4.2.1.1 Market-Related Factors: 

The competitive forces in the airfreight market cause this group of hinders. These are: 

1. Emergence of integrators: The integrators, although traditionally had been dealing with 
documents and small packets, have diversified their service and have achieved a significant 
share of the market. This has deteriorated belly utilization. 

2. Competition in short destination flights: The short destination flights contain less freight than 
the long destination ones since it costs the customers much more to send the freight bv air 



than by truck. If logistically the freight is to be transported fast according to the customer, 
only then the airlines are asiced for the service in short distant flights. Consequently, for the 
short destination flights there used to be a huge empty capacity. 



4.2.1.2 Strategic Factors: 

Different strategies that airlines have in their operations cause this group of hindering factors. 

These are: 

1. Priority maintenance: There are many activities that take place during the turnaround process 
for an aircraft. The passengers get the highest priority to be boarded into the aircraft. Then 
comes the baggage of the boarded passengers, the post and then freight with the lowest 
priority. This means that if there is not enough time for the freight to get the flight, the freight 
is left back at the airport. 

2. Smaller capacity aircraft: Type of aircraft is very important for a good capacity in the belly. 
Wide-body aircraft is freight friendlier than slim-body aircraft. The demand of the industries 
is, generally, shipment of bigger volume freight. For the demanded volume of shipment, for 
the distribution and logistics it is very difficult to reach a solution based on capacity in the 
slim-body aircraft. Also, if the passenger volume is not much then generally an even smaller 
type of airline is used, which is even worse for the huge amount of freight. 

3. Use of belly freight only for express delivery: Different airlines have different strategies with 
the pricing of the belly airfreight. For some airlines belly is an express freight and should be 
paid accordingly, i.e. a higher price. They do not want to ship normal freight through the 
belly even if there is empty capacity. 

4. Separate organizations: For some airlines the cargo and passenger units are separate 
organizations and the cooperation between them is often poor resulting a worse business. The 
passenger unit wants the best of the passengers and does not want to think about the benefit 
of the cargo unit if that collides with the benefit of the passenger unit. When, for example, it 
concerns purchasing of new aircraft they prioritize issues like speed, environmental- 
friendliness of the vehicle and even cost for the runway while the weight of the aircraft 
exceeds a certain capacity. 

5. Fueling: Sometimes the aircraft takes fuel in a place where the price is cheaper. This 
increases the weight of the airlines that results less freight in the belly. Also, for longer 
destinations the aircraft must contain a huge amount of fuel, which also decreases capacity in 
the belly. 

6. Capacity problem in HUB: The HUB might have limited capacity and cannot process goods 
when it exceeds a certain amount. This might cause the flights not to exceed certain load so 
that the ultimate pressure in the hub does not exceed it's capacity. 

7. Prioritizing permanent customers: There is often a problem with the permanent customers 
that they deliver more (or less) than they are supposed to. The deviation is not informed 
earlier to the airline which makes it difficult to plan the load. If the arrived goods are much 
more than informed, others' freight is unloaded prioritizing the goods of the permanent 
customers. These customers even send less goods than they are supposed to. In that case the 
belly utilization is less because by that time they had already refused to take goods from 
other customers assuming that the permanent customers would send the promised amount of 
goods. 



8. Concentration on pure freighter. As the airfreight market is increasing very fast some of the 
airlines using pure freighters to tackle the demand. Thus, dependency on belly goes down 
which results in no concentration in planning of a better utilization of it. Passenger 
prioritization for these airlines rises even higher, which results in a less belly capacity. 

9. Less cooperation with the customers: The forwarders generally do not let the airlines to be 
involved in the relation between the forwarders and their customers. Information that could 
directly be conveyed to/from the customers goes via forwarders instead and thus it takes 
longer time to solve the problem. 

10. Deadline to deliver freight at terminal: Deadline of delivering the goods for belly traffic 
might vary between airlines. Sometimes the time (demanded by the airlines) for operations 
on the goods at the terminal might be more than necessary. If the deadline (for customers) to 
deliver the goods at terminal is increased, the airline could get some more customers and thus 
process more freight for the belly. 



4.2.1.3 Delivery-Related Factors: 

This group of hinders are caused by the customers or the forwarders while delivering goods to 

the airport. These are: 

1. Goods arriving the terminal mostly at the same time: The goods arrive at the airport mostly 
in the evening. The airplanes, on the other hand, run all day long. This points out to an 
uneven utilization of belly capacity. Moreover, to some destinations it is too late to ship the 
goods in the evening since the goods cannot be custom-checked in the destination airport and 
thus can not reach the customers in the same evening. (Moreover, the fact that most of the 
goods arrive at the same time can mean a quality problem in the handling of the goods.) 

2. Late arrival of the goods at the airport: Late arrival of the goods at the airport may cause 
planned shipment cancelled. 

3. Wrong information about the shape/weight of goods: It might be a problem to load the goods 
if incorrect information about its shape or weight is received. Belly has shape and weight 
constraints. If the information about the weight or shape is wrong, it may lead the plane to 
leave the airport without carrying the goods. 



4.2.2 Factor Group 2 

Factor Group 2' represents hindering factors that are caused by operational inadequacies at the 
airport, more precisely, at the apron. This group contains a single subgroup of factors. This 
subgroup is termed as 'operational' factors. 

4.2.2.1 Operational Factors: 

This group of hinders are caused by inefficient operation on the goods after the goods are 

delivered at the airport until loaded into the aircraft. These are: 

1. Congestion at ramp at the peak hours: Sometimes congestion at ramp (caused by different 
servicing vehicles at the ramp) at the peak hours constrains the loading possibility. 



2. Weight limitations of handling equipment at the ramp: Sometimes the mechanical equipment 
cannot handle weights of certain capacity for the belly cargo to be utilized. Even if the source 
airport is equipped, the destination airport might have insufficiency to handle certain amount 
of weight. 



4.2.3 Factor Group 3 

Factors described in "Factor Group 3' are related to problems around the aircraft. Hindering 
factors in this group are caused by aircraft unavailability and also incompatibility of aircraft 
capacity with the amount of goods to be loaded. This capacity minimization could be due to less 
time for loading the goods or due to less physical capacity in the aircraft resulting whole or part 
of planned shipment cancelled. 

4.2.3.1 UnavailabiUty Issues: 

Belly utilization is decreased through unavailability of the aircraft in the terminal. The following 
factors can cause aircraft unavailability: 

1. Cancellation of the aircraft: Cancellation of flights due to technical or other problems makes 
planned shipment often cancelled. It is easier to mange with the passengers rerouting their 
journey. Because the transfer time (in a third airport) is much less for the passenger than for 
the freight. 

2. Non-arrival of scheduled flight: Non-arrival of scheduled flights (because of cancellations in 
the source airport) makes planned shipment cancelled. 

4.2.3.2 Incompatibility Issues: 

Even if the aircraft is available, the aircraft capacity might be incompatible with the amount of 
goods to be loaded. Incompatibility might be generated due to less loading time or less physical 
capacity in the aircraft. The following factors may cause incompatibility: 



1. 



2. 



J. 



Late arrival of the aircraft: Late arrival of the aircraft makes the turnaround time shorter. The 
aircraft turnaround time is very important to maintain because it plays a vital role for the 
image of the airlines. If an incoming flight arrive late at the airport, scheduled loading of 
freight might be cancelled or the amount allowed might be less in order to maintain a fast 
turnaround. Certain type of goods (big pallets, for example) is hard to load in a shorter 
turnaround time. 

Unexpected amount of passenger baggage: Amount of passenger baggage may vary because 
of unanticipated amount of passengers. Even with the same amount of passengers, the 
baggage amount may vary. In a slim-body passenger aircraft Generally 1000 to 3000 kg's of 
freight is possible to load after loading the baggage of the passengers. But sometimes this 
estimation does not work because of uncertain amount of luggage enter the belly leaving a 
few or almost no capacity to be used by the freight. Preventive measures could be taken in 
order to increase the probability of utilization of the belly. 

Sudden change of aircraft: If number of passengers is less than anticipated the planned 
aircraft may be changed to a smaller type. This might cancel planned shipment. 



4. Weather conditions minimizing lifting capacity: Weather conditions (like headwind) in 
specific geographical locations sometimes limit the aircraft to exit certain weight. The belly 
gets affected since it has the last priority to get the aircraft. 



5 ANALYSES OF FACTORS 

In this section, the impact of the hindering factors (that are identified through interviews with the 
different actors) on the whole process of belly utilization will be analyzed. 



Factor Group 


Number of hinders contained 


% of total 


1 


15 


65 


2 


2 


9 


3 


6 


26 



Table 2: Share of factor groups causing less belly utilization 

The hindering factors in FGl (Factor Group 1) minimize the possibility of delivering the freight 
to the terminal (figure 3). Which means, the less the strength of the factors in this group the 
better it is for the availability of freight at the airport. The hindering factors in FG2 complicate 
the chances of the existing freight to be transported to the belly and thus minimize the belly 
utilization. Factors in FG3 minimize it even further. The less the strength of the underlying 
hindering factors (i.e. the less the possibility that a certain constraining factor will occur) the 
more the possibility of a better belly utilization. FGl (with the underlying factor groups — 
market related factors, strategic factors and delivery-related factors) appears to be the strongest 
(at least in quantity of hindering factors) among all the groups. FG2 contains the lowest number 
of factors. If the goods flow will be much more (that is if influence from FGl is less) than new 
operational hinders and also strengthening of mentioned hinders might occur weakening the 
belly utilization. More goods flow might even strengthen the factors belonging to FG3. 

Even if more hindering factors might be generated or strengthening of the existing hinders might 
occur, these should not prevent more goods flow at the airport. Along with the generation of 
hinders, preventive measures should also be taken and implemented. 

As we can see in table 2, FGl contains 65% of the all the hindering factors. FG2 and FG3 
contain 9% and 26% of the factors respectively. If we consider that each of all the 23 factors 
described above weigh the same, FGl obviously draws the concentration. The second argument 
to concentrate mostly on FGl is that there must be goods at the terminal in order to process it all 
the way to the aircraft. 



6 ACTOR VERIFICATION 



In this section, the estimated strength of the factor groups analyzed in the previous section will 
be verified with the quantitative data collected through a questionnaire survey. The survey 
includes three main commercial airports in Sweden, which are Stockholm-Arlanda, Goteborg- 



Landvetter, and Malmo-Sturup. For each of the airports, main actors (or airlines) utilizing 
passenger aircraft belly for freight flow were defined. The different ground-handling companies 
operating in the respective airports were contacted in order to know who the main actors were. 
After being informed about the main actors, they were contacted and people best suited to the 
goal of the survey were searched for. After being convinced that the right person (to answer to 
the survey) for each of the airlines was found, the questionnaire was send to the respective 
person. The total number of questionnaire sent to the actors was 20. To be mentioned that for a 
single airline one individual answered to the survey representing the respective airline's 
operations in two airports. Thus, the data received in a single questionnaire represents 
experiences of two different airports (i.e. one additional airport). There was also a case where a 
single airline was responsible for freight operations for two additional airlines. In this case, the 
data received in a single questionnaire represents three airlines. This means that although the 
total number of questionnaire sent to the actors was 20, they were responsible for data on more 
than 20 operations. 12 of the 20 questionnaire were answered, which denotes a response of 60%. 
Taking into account that one person answered for a single airline's operation in different airports 
and a single airline answered also for different other airlines, 15 (i.e. 1+2+12) of the 20 
questionnaire were answered denoting a response of 75%. 

The airlines surveyed were - KLM, SAS, British Airways, Air France, Finnair, Thai Airways, 
Swissair, Sabena, Austrian, Aeroflot, Premiair, Malmo Aviation, Novair, Iberia, and Delta 
Airlines. 

Two questions were mainly focussed in the survey. The first one was to estimate the degree of 
influence each of the hindering factors had on the operations of the actors. The second one was 
to estimate how the actors considered the possibility to eliminate the hindering factors. Figure 4 
demonstrates the influence of the factors and the possibility to eliminate the factors as received 
through the survey. 



50 
45 
40 
35 
30 
25 
20 
15 
10 
5 




1 



U 



□ Degree of Influence 
■ Elimination Possibility 



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 



Hindering factors 

Figure 4: The degree of influence of different hindering factors and their possible elimination 

To be mentioned that all the respondents answered to the first question and only one of the 
received questionnaire did not have any answer on the second question. In order to have a good 



comparison, data on the second question needed to be adjusted'. Table 3 shows survey data on 
both the questions including the adjusted data for the second question. 



Hindering Tactors 


Degree of 
Influence 

(A) 


Elimination 

Possibility 

(original) (B) 


Elimination 

Possibility 

(adjusted) (C) 


Difference 
(A-C) 


1. Emergence of integrators 


32 


29 


32 





2. Competition in short 
destination flights 


38 


28 


31 


7 


3. Priority maintenance 


35 


26 


28 


7 


4. Smaller capacity aircraft 


47 


23 


25 


22 


5. Use of belly freight only for 
express delivery 


18 


38 


41 


-23 


6. Separate organizations 


33 


27 


29 


4 


7. Fueling 


19 


21 


23 


-4 


8. Capacity problem in HUB 


24 


33 


36 


-12 


9. Prioritizing permanent 
customers 


37 


30 


33 


4 


10. Concentration on pure 
freighter 


27 


29 


35 


-8 


1 1 . Less cooperation with the 
customers 


39 


32 ■ 


35 


4 


12. Deadline to deliver freight 
at terminal 


27 


31 


34 


-7 


13. Goods arriving the terminal 
mostly at the same time 


38 


19 


21 


17 


14. Late arrival of the goods at 
the airport 


29 


27 


29 





15. Wrong information about 
the shape/weight of goods 


31 


31 


37 


-6 


16. Congestion at ramp at the 
peak hours 


26 


28 


34 


-8 


17. Weight limitations of 
handling equipment at the ramp 


25 


28 


31 


-6 


18. Cancellation of the aircraft 


33 


9 


10 


23 


19. Non-arrival of scheduled 
flight 


30 


10 


11 


19 


20. Late arrival of the aircraft 


37 


16 


17 


20 


21. Unexpected amount of 
passenger baggage 


39 


17 


19 


20 


22. Sudden change of aircraft 


35 


13 


14 


21 


23. Weather conditions 
minimizing lifting capacity 


25 


12 


13 


12 



Table 3: Surveyed strength of the different hindering factors and their possible elimination 



If we accumulate the surveyed strength (table 3) of all the hindering factors according to the 
three factor groups, we can analyze the hindering strength that each factor group has in the 
operation. 



C =B>:- 



12 



Factor 

Group 

(FG) 


Itof 
underlying 
factors in 

FG 


Sub-factors 


Accumulated 

strength of 

underlying 

factors 


Accumulated 

strength of 

FG 


Average 
strength of 
each hinder 


1 


15 


Market-related factors 


70 


474 


31.6 


Strategic factors 


306 


Delivery-related factors 


98 


2 


2 


Operational factors 


51 


51 


25.5 


3 


6 


Unavailability issues 


63 


199 


33.1 


incompatibility issues 


136 



Table 4: Hindering strength of each factor group 

As we can see in table 4, FGl has the biggest influence among all the factor groups with an 
average strength of 31.6 for each underlying hinders. Although this average is less than the 
average hindering strength of hinders in FG3, the superiority in total number hinders (in 
comparison to that of FG3) makes the dominance of FGL On the other hand, if we compare the 
hindering strength with the possibility of their elimination, we can see (figure 5) that elimination 
of the factors underlying in FGl is relatively easier than elimination of factors in FG3. 



Factor Group 1 
I 



Factor Group 2 Factor Group 3 
I I 



30 
20 
10 


-10 
-20 
-30 



22 



m 



1 ^ T 



-4' 

m 






r^,„ R 






73 14 



w 



If i4 iS 20 Ji 




■12 



■e 



-23 



Figure 5: Differences between hindering strength and eliminating strength 

It is obvious that in order to generate a better goods flow at the airport, it is necessary to reduce 
the strength of factors in FGl. Also, according to the data, it is relatively easier to confront this 
group of hinders. If, after taking care of strategic or other changes, the strength of FGl is reduced 
creating a better goods flow at the airport, this will increase hindering strength of factors in both 
FG2 and FG3 provided the underlying hinders are not reduced. If (in relation to a better freight 
flow) the operational inefficiency at the ramp is not increased or the required capacity is not 
compatible with the amount of goods, the reduction of hindering strength in FGl would mean 
nothing. This necessitates concentration on FG2 and FG3. According to figure 5, it seems that 
the actors are able to have better control on the factors in FG2. When it concerns FG3, the ability 
to control the underlying factors seem not to be satisfactory. 

7 CONCLUSIONS 

As the aircraft-turnaround time is limited the cargo is de-prioritized enters the aircraft latest in 
the process. In the study, we have found that for a planned belly shipment to be made, the aircraft 



12 



must stay at the ramp no less than the duration of the planned turnaround time. This necessitates 
the aircraft to land on time. The different activities in the turnaround should also be performed 
effectively enough so that the time needed to load the cargo is available. Other deviations (e.g. 
change of aircraft to a smaller size etc.), that can minimize the capacity for cargo in the aircraft 
should not exist. The chances that deviations take place are not rare. Customers, that want to ship 
goods by air need fast transportation and pay relatively higher than they would have if 
transported by any other mode. In relation to the price that customers pay, the service level must 
be satisfactory. But as the different hindering factors worsen the chances of a shipment, this 
might make it harder for the operators to motivate themselves to think of making customer 
relations based on the empty belly utilization in passenger aircraft. Improved control over the 
hindering factors, especially those related to the apron and the aircraft (i.e. FG2 and FG3) can 
motivate the actors to make better customer relations and promote possibility to utilizing the 
belly more efficiently as chances to more satisfactorily serve the customers are higher. These 
factors hinder flow of goods in the airport, i.e. from the goods terminal to the aircraft belly. 
Hindering factors that exist outside the airport area (i.e. FGl), hinder goods flow from the 
customers to the airport. Before reducing the impact of FG2 and FG3, reduction of the impact of 
FGl would mean more goods at the airport creating more pressure with the limited capacity. 
That is why we stress on improvement especially in FG2 and FG3. It is clear from the survey that 
the underlying factors in these two groups are not easy to tackle. Factors related to unavailability 
issues, like cancellation or non-arrival of scheduled flights are very much problematic since it is 
hard to ship the goods through connecting routes (i.e. as done with the passengers). Preventive 
measures should be taken for both the unavailability and the incompatibility issues in order to 
improve the underlying hinders as much as possible. The improvement will assure more goods to 
be shipped, the actors then can concentrate on reducing factors underlying in FGl in order to 
have a better goods flow to for the belly to be utilized. 

The aim of this paper was also to analyze if a more efficient utilization of the belly at the daytime 
can generate a new 'overday' concept of processing airfreight and thus achieve a more significant 
share of the market. Integrators, as we know, have a concept of 'overnight' delivery of goods and 
they are quite successful with this concept triggered by the fact that they have the goods 
available at the airport in the evening for shipment. They own all the assets (e.g. different 
vehicles, information system, etc. that are necessary) in the whole process of their services. 
Consequently, they have a good control over the freight flow and can effectively deliver the 
goods. Belly's flexibility, on the other hand, lies with the frequency of the flights. But, as most of 
the goods arrive at the terminal in the evening, this frequency does not mean a lot for the 
operators. The result shows also that the control is not satisfactory over the hinder. In order to 
look forward to generating an 'overday' concept, hinders underlying in all the factor groups 
should be minimized as much as possible so that more freight flow is generated at the airport and 
satisfactory service is provided. The airlines have to realize that their service has contributed to 
build up the customers' 'trust'. As day-time is not an appropriate hour for the shippers to deliver 
the goods at the terminal, tempting propositions should be made by the airlines so that the 
customers are interested to deliver goods all over the day. It needs further research to judge 
whether it is possible to generate a 'overday' concept of passenger aircraft belly utilization as it 
depends on how much of the hindering impact the airlines are willing to minimize. It depends 
also on how closely the actors then (with reduced hindering impact) cooperate with each other 
and e.xploit the advantages and disadvantages through such a process. 



8 REFERENCES 

Acharjee, Prabal and Lumsden, Kenth, Aircraft turnaround study for Goteborg/Landvetter 
airport. Department of Transportation and Logistics, Chalmers University of Technolo<^v 
Goteborg, Sweden 1999 °^' 

Arbnor, Ingeman and Bjerke, Bjora, Methodology for Creating Business Knowledge, Second 
Edition, Studentlitteratur, Lund, Sweden 1994 

Dahllof, G., "Konservatism hotar flygbolag och speditorer", Transport och Hanterins TH 
Forlag, Sweden, No 1-2, 1997 

Larson L-G., PMsson A., Tekniska mojligheter att minska avgasutsldppen fr&n flvstrafik i 
5ver/ge,FFATN 1996-33, Stockholm 1996 ^ j j's J 

Lumsden, Kenth, Logistikens Grunder, Teknisk Logistik, Studentlitteratur, Lund, Sweden 1998 
Transport och Hantering "1997 ett oerhort bra flygfraktsSr", TH Forlag, Sweden, No 12, 1997 



14 



Re-examining the Slot Allocation Problem 



Vanessa Tamms 

Australia- Japan Research Centre 

Asia-Pacific School of Economics and Management 

Australian National University 

Ph: +612 6249 5532 

Fax: +61 2 6249 0767 

Email: vanessa.tainms@anu.edu.au 



Paper presented at the Air Transport Research Group Conference, 
City University of Hong Kong, 6-8 June 1999 



Abstract 

This paper puts forward the case for re-examining the feasibility of using 
auctions to allocate take-off and landing slots at airports in light of the success of the 
US radio spectrum auctions. It discusses how the simultaneous multi-round design of 
the spectrum auctions would need to be adapted to accommodate combinatorial and 
contingency bidding behaviour, given the synergies inherent in operating particular 
combinations of slots and the substitutability of slots within certain time intervals. It 
also highlights how broad cooperation would be required to implement such a system 
across airports located in many different countries. Finally, it suggests that the right to 
provide services also be embodied in the definition of a slot in order to ensure that 
auction outcomes are efficient. 



Introduction 



The past decade has witnessed liberalisation of international air transport 
markets. Many countries have substantially deregulated their domestic markets, 
privatised their national "flag carriers", and permitted multiple designation of carriers 
on international routes. Many have also entered into agreements with other countries 
which seek to liberalise markets between themselves and the partner country. The 
European Economic Area' (EEA) countries, for example, have established a Common 
Aviation Market (CAM) by replacing the bilateral Air Services Agreements (ASAs) 
which formally governed trade in air transport services among them with a 
multilateral agreement. The US has so far replaced 33 of its ASAs with "open skies" 
agreements (liberalised bilateral agreements), and Australia and New Zealand have 
established a Single Aviation Market (SAM) across the Tasman. 

Despite these moves, significant barriers to trade in air transport services 
remain. Foreign investment remains highly restricted, and hence the liberal terms of 
the replacement agreements generally only apply to carriers registered in signatory 
countries. The provision of cabotage services also remains prohibited, except for 
under the CAM and SAM agreements which permit this by carriers based in signatory 
countries only. These restrictions have not only prevented third-country carriers from 
providing services in particular markets, but in most cases also carriers from 
providing domestic services in a foreign country. They have also restricted carrier 
mode of supply to "production" in the country of registration and "export" abroad. 

In addition, airport- and ticket-sales-related issues have received little attention 
from regulators, with two exceptions. The EEA has issued Council Directive 
96/67/EC on access to ground-handling services, which permits EEA carriers to self- 
handle land-side at all EEA airports, and self-handle airside at all EEA airports with 1 
million passengers and/or 25 000 tonnes of freight or more per annum. Ground- 
handling by third parties is also currently being phased inl The General Agreement 
on Trade in Services (GATS) includes an Annex on Air Transport Services, which 
seeks to make aircraft repair and maintenance, travel agent and computer reservation 
system (CRS) ser\'ices consistent with the transparency, non-discrimination and 
national treatment rules of the worid trading system. However, the majority of GATS 
signatory countries were granted exemptions from the three Annex provisions, such 
that at present the Aimex is virtually ineffective^ 

This paper focuses on the allocation of take-off and landing slots at major 
airports, one of the airport-related issues which has received virtually no attention 
from regulators. It describes methods currently used to allocate slots, and explains 
how these not only make significant competitive new entry difficult, but also prevent 
incumbents from operating efficient networks. It also puts forwards the case forre- 



^ The 15 Member States of the European Union plus Norway, Iceland and Liechtenstein. 

- This is scheduled to be in place at all EEA airports with 2 million passengers or 50 000 tonnes of 

freight or more per annum by 1 January 2001 (Official Journal of the European Communities No L 

272. 25/10.96; 36). 

■' A new round of negotiations is scheduled to begm in 2000, at which lime these exemptions will 

expire. 



examining the feasibility of using auctions to allocate slots, given the success of the 
US radio spectrum auctions and the similarities between the tasks of allocating 
spectrum and slots. It also highlights how the design of the spectrum allocation 
mechanism would need to be adapted to accommodate combinatorial and contingency 
bids given the complementarities among, and substitutability of, airport slots, and how 
broad cooperation would be required to implement such an allocation system across 
the world's major airports, given that these are located in many different countries. 
Finally, it puts forward the case for embodying the right to provide services as well as 
the ability to take-off and land in a slot as in the case of radio spectrum, reducing the 
current two-step procedure to a single step and improving efficiency. 



Section I: Methods Used to Allocate Slots at Major Airports 



The lATA System 

Historically slots were largely allocated on a first-come first-served basis. As 
air traffic grew, however, airport congestion grew and so did delays. Airlines 
subsequently established scheduling committees at major airports which aimed to 
better coordinate take-offs and landings such that delays would be minimised. By 
1993 there were over 100 of these committees in operation around the world". 
Traditionally the committee at each airport consisted of staff on secondment fi-om the 
major incumbent airline serving that airport. However, it is now a requirement that at 
fully coordinated airports (airports at which demand is greater than supply at most 
times of the day) a panel consisting of the carriers which are the largest providers of 
services at these airports oversee the process^. Before each season commences, the 
airport authority, on advice fi-om air traffic control, determines the feasible number of 
aircraft movements (take-offs and landings) at each hour of the day. Carriers currently 
serving or wishing to serve a particular airport submit their slot requests to the 
scheduling committee of that airport, and the committee allocates slots among carriers 
based on their requests and the feasible number of movements. 

At times of the day where demand for slots outstrips supply, committees 
allocate slots according to rules set out by the International Air Transport Associafion 
(lATA), the trade association for airhnes. These rules give priority to carriers which 
request slots they used in a previous equivalent season (summer or winter), then 
carriers wishing to change the times of exisfing services, new entrants, carriers 
wanting to extend existing services to year-round operations, and then carriers whose 
schedule will be effective for a longer period of operation in a particular season . 
Since 1990 lATA rules have required that 50% of unclaimed slots, slots that become 
newly available, or those surrendered under the "use-it-or-lose it" rule in each time 



* Jones, Viehoff and Marks (1993); 40 

' The chair of the panel is still generally a former employee of the major incumbent airline: at London/ 

Heathrow, for example, the chair is a former British Airways employee. 

^ If slots are still available and there are requests for these, secondar>' criteria are used to allocate them. 

These include: the need for a mi.\ture of long-haul and short-haul operations at major airports, the 

effect on competition, consideration of curfews at other airports, and requirements of the travelling 

public and other users iL\TA (199S), Scheduling Procedures Guide, Twent>--Thu:d Edition, Januar>'; 

10). 



period be allocated to new entrants, where new entrants are defined as those carriers 
which hold no more than four slots per day at the airport in question. The lATA rules 
have basically been accepted into European Union (EU) law, with the exception that 
the definition of new entrant has been extended to airlines seeking to provide 
competition on intra-EEA monopoly and duopoly routes which hold fewer than four 
slots a day for that service, provided that they are seeking no more than a twice daily 
service, under Council Regulation 95/93 adopted in 1993. Any airline with more than 
3% of all slots at an airport or more than 2% of slots at an airport system cannot 
qualify as a new entrant^. 

Given that each airport scheduling committee makes its decisions 
independently, lATA Schedule Coordination Conferences are held bi-annually to 
enable airlines to coordinate their schedules worldwide. At these conferences, carriers 
are able to swap slots with others under antitrust immunity, in order to try to obtain 
slots they still require or consistent sets of arrival and departure times. They may trade 
slots at different airports and alter the type of aircraft flown, subject to the approval of 
the relevant scheduling committees. However, no money may change hands, which 
means that often trades need to involve many parties simultaneously, making the task 
complex and time-consuming. Trading can also take place after the conference on an 
ad hoc basis*. 

The priority given to "grandfathering" means that the majority of slots at 
congested airports, particularly during peak periods, are retained each season by 
incumbents. The few slots (if any) which are available will tend to at non-peak times 
and inconsistent across days of the week. This has severe consequences on 
competition and efficiency given the nature of passenger demand and the economics 
of providing air transport services. Studies have shown that passengers, particularly 
time-sensitive passengers, prefer frequent services on short-haul routes: airline yield 
increases more than proportionally the greater the number of daily frequencies a 
carrier offers on a particular route as business passenger numbers increase more than 
proportionally . Passengers also prefer interlined consecutive services, as collecting 
baggage and re-checking in is not necessary on these at intermediate stops. They also 
show a preference for carriers on whose flights they can accumulate frequent-flyer 
points . There is also substantial evidence that there are significant economies of 
traffic density inherent in the provision of air transport services: an increase in 
network traffic, via an increase in flight frequency per route, greater average load 
factor, or through consolidating passengers onto larger aircraft will decrease a 
carrier's average costs". 

Carriers will thus aim to provide high-frequency, well-interlined services on 
which passengers can earn frequent-flyer points in order to satisfy passenger demand 
and achieve the maximum cost-savings inherent in the provision of their services. 
Under the lATA system of slot allocation, however, both new entrants and 



^ Official Journal of the European Communities No L 14, 22/1/93. 
' Jones, Viehoff and Marks (1993); 41. 



This phenomenon is known as the ■"origin-point-presence effect" or the "s-cun'e effect" (Tretheway 
and Oum (1992)). 

' Morrison and Winston ( 19S9). usiin 19S3 US data, found that on average passengers were willing to 
pay an additional USS32 per round trip in order to accumulate frequent flyer pomts. , 
" Caves, Christensen and Tretheway (19S4). 



incumbents are restricted in their ability to do this. New entrants are unHkely to be 
allocated sufficient quantities of peak-period slots consistently timed across the week 
which would enable them to provide high-frequency, well-interlined services and 
hence compete effectively with incumbent carriers. Experience in the EEA shows that 
carriers which have been allocated slots from the 50% of the slot pool reserved for 
new entrants have not subsequently begun providing services on short-haul routes in 
competition with incumbents; rather, they have either handed them back, used them to 
increase frequency of service on routes they already serve, or begun providing long- 
haul services'^. 

The prohibition of slot sales further hinders new carriers from entering 
markets. As already mentioned, the lATA system only permits slot swaps, and hence 
carriers must initially have something to swap in order to take part in this process. 
Slot sales allegedly take place in post-conference trading, disguised as slot swaps: 
carriers exchange slots, where this is accompanied by an under-the-table financial 
payment from the holder of the slot with the lower market value to the holder of the 
slot which has a higher value. Even carriers which have not been allocated slots at the 
airport in question can take part in this by applying for an off-peak slot, at say Sam 
and then "swapping" this for the slot they desire. The slot obtained by the other carrier 
is subsequently returned to slot pool under the "use-it-or-lose-it" rule of the LATA 
system. However, it is unlikely that a new entrant will be able to acquire either the 
number of the type of slots it would require to establish viable services on short-haul 
routes, especially when some of the potential trading partners are carriers which it 
would compete directly with should it be able to acquire them. Even if slots allocated 
could be bought and sold, however, new entrants would be at a disadvantage as they 
would be forced to pay for something which incumbents were initially allocated free 
of charge. 

Obtaining the necessary slots is likely to become more of a problem over time 
given the increasing number of airlines establishing inter-carrier alliance agreements, 
as not only will the pool of sellers shrink, but also new entrants will have to provide 
more frequencies in markets where the partner carriers both provide services and 
coordinate their schedules. New entrants could themselves enter into code-sharing or 
block-purchasing agreements with carriers already serving the markets in which they 
wish to operate. However, this only allows them to indirectly serve these markets. In 
addition, their ability to do this will also depend on the success they have in finding a 
partner, which in turn will depend on the extent to which incumbents already have 
such agreements with the new entrant's competitors. 

A carrier may be able to begin providing services in particular markets if a 
second airport exists in a city which can accommodate the type of services also 
operated out of the first and it itself is not slot-constrained at peak periods. However, 
these airports are often not perfect substitutes for one another: time-sensitive 
passengers in particular often prefer one airport over another due to locational or other 
factors. In London, for example, many passengers show a distinct preference for 
Heathrow Airport compared to Gatwick, given the former's closer proximity to the 
centre of London and its greater number of connection possibilities'" . These airports 



'- UK Civil Aviation Authont>- (199S) 



'■' GRA Inc (1996) state that yields on Heathrow sen'ices are of the order of 10'\. hi^her.than 
equivalent Gatwick ser\'ices. 



will become closer substitutes the longer the flight-length or the greater the price 
differential between the same flight operating out of the primary and secondary 
aiiports. Experiences in the US and the UK show that secondary airports are 
extremely effective substitutes if the price differential is large enough: Southwest 
Airlines and easyJet operate out of Dallas/ Lovefield and London/ Luton 
respectively . Indeed, carriers can even viably provide low-frequency, non-interlined 
services which do not provide frequent-flyer benefits at non-peak times of the day if 
airfares are sufficiently low. 

It is not necessarily the case that new-entry is more beneficial than an increase 
in services by incumbent carriers. Indeed, the smaller the number of slots obtained by 
the entrant, the more likely it is that incumbents could have used these more 
effectively. For example, an incumbent may use them together with slots it already 
has to begin providing services in new markets in competition with other carriers, 
which not only leads to a reduction in airfares in these markets, but also improves the 
connectivity of its network and offers its passengers more ways to earn frequent-flyer 
points and redeem accumulated mileage. Alternatively, it may use them to begin 
providing services in a market not currently served, which may provide greater 
benefits to passengers than through having a new competitor in a market ah-eady 
served. Even if the new entrant uses the slots in the same way in which the incumbent 
way planning to use them, it may be the case that the incumbent is more efficient than 
its new entrant counterpart, and hence that additional inefficiencies are being 
introduced into the market. 

The difficulties experienced in obtaining slots by would-be new entrants 
protects incumbent carriers fi-om competition, enabling them to price their services 
above the long-run competitive level. However, this must be traded-off against the 
fact that they are similariy unlikely to be allocated slots which would enable them to 
add to their existing route networks. Indeed, they may even be unable to obtain the 
necessary slots to add frequencies to routes they already serve in order to more fully 
capture economies of density given that if they apply for new slots their request 
receives lower priority than those of new entrants. In order to begin operating new 
routes or increase flight frequency, incumbents must thus sacrifice services they 
currently provide such that the slots they require become available. This limits the 
extent to which carriers can respond to passenger demand and capture savings from 
economies of traffic density. 

An additional problem with the lATA system is that it is not internationally 
binding. In practice this means that allocation is prone to intervention by national 
governments reacting to political pressures, and that there is little adversely affected 
carriers can do about it except lodge complaints. A recent example of this occurring 
was the allocation of "new" slots at Tokyo/ Narita Airport in the second half of 
1998 ^ When this occurs, the system is not consistent with the non-discrimination, 
national treatment and transparency principles of the v/orld trading system. 



Lovefield Aiiport is within closer proximity of the centre of Dallas than Fort Worth Airport; 
however, the latter has better connection possibilities. 

' 202 "'new" slots became available after the conclusion of a Memorandum of Understanding between 
Japan and the US; these were subsequently mainly awarded to Japanese and US carriers, most of which 
already have substantial presence at this airport. US carriers claim that these slots became available 
because a number of (unused) slots were surrendered by Federal E.xpress and because the Japanese air 



Finally, governments forego revenue from carriers which would normally 
accrue to asset-holders. This revenue could be used to improve the budget bottom line 
or provide tax-relief to corporations or individuals. 

The US System 

The lATA system cannot be used to allocate slots at US airports'^, as under 
US anti-trust laws US carriers cannot meet to discuss flight scheduling. Airlines 
simply schedule their flights taking into account expected delays at the busier 
airports'^, except at the designated "high-density" airports: Chicago/ O'Hare, 
Washington/ Reagan National, New York/ John F Kennedy and New York/ La 
Guardia. Slots used for domestic services at these four airports have been subject to 
different rules since the introduction of the "High Density Traffic Airports Rule" by 
the FAA in April 1969. The rule established slot quotas for scheduled air carrier 
services, commuter services and general aviation at these airports. 

Initially scheduling committees were set up at each of these four airports, 
where each consisted of committee staff and carriers serving or wanting to serve the 
airport in question. Unlike the situation at flilly-coordinated airports which abide by 
the lATA rules, however, after all slot requests were received all members of a 
scheduling committee would meet together and multilaterally negotiate the 
withdrawal of requests for slots until the number sought equaled the number available 
at all times of the day. These meetings were granted anti-trust immunity; however, 
post-corrmiittee meeting gatherings to coordinate schedules across airports were not 
permitted. Importantly, any distribution of slots was required to be unanimously 
agreed upon by committee members; if agreement was not forthcoming within a 
certain time period the responsibility of allocating slots would be handed over to the 
FAA. The rules the FAA would use to allocate slots in such a situation, however, 
were unknown'*. Grether, Isaac and Plott (1989) used controlled environment 
experiments to show that in such circumstances committee decisions will tend to be 
substantially governed by the perceived consequences of default. Larger carriers 
apparently thought that the FAA would grant new entrants at least a small number of 
slots, and hence "conceded" these in conmiittee meetings in order to avoid default. 

Initially the system encouraged new entry because potential entrants knew that 
they were almost guaranteed to gain some slots given their ability to cause the 
committee to default, but the scale of new entry was generally small. Many of the 



traffic controller's union agreed to raise hourly traffic movements. However, many of the slots which 

became available were at different times to the surrendered slots. There is industry speculation that 

these slots were "found" in order to appease the US given that the new (second) runway is unlikely to 

be fully operational for several years, and that the compliance of Japanese carriers was obtained in 

return for assurances that they would be looked upon favourably when the slots which become 

available prior to the opening of the new runway are allocated. The Europeans subsequently formally 

complained and threatened sanctions against Japan (Airline Business, August 199S; 28-29 and October 

!99S;26). 

'* -Agencies (usually the Federal Aviation .Administration (F.A.A)) represent US earners in LATA 

schedule coordination conferences uhich involve the trading of slots used for international sers'ices at 

US airports (Starkie (1992); 27). 

'"Starkie(199S); 113 

'^ Grether, Isaac and Plott (19S9), 



outcomes of the system were similar to those of the lATA system: new entrants were 
not necessarily more efficient than incumbent operators, incumbents were prevented 
from expanding their services in line with passenger demand without sacrificing 
existing services and hence were preventing from fully capturing economies of traffic 
density inherent in the provision of their services. However, in addition, the system 
made it difficult for carriers to coordinate system-wide operations given that the 
decisions each committee were made independently of allocations at other airports. 

Deregulation of US domestic aviation markets in 1978, however, led to 
increased demand for slots, making it more difficult for scheduling committees to 
reach consensus. The FAA was eventually forced to intervene in 1980 when new 
entrant New York Air sought a large number of slots in order to establish low-cost 
services between Washington/ Reagan National and New York, and it took slots from 
incumbents for redistribution among new entrants'^. The system was suspended in 
response to the air traffic controllers' strike in 1981, and in 1982 carriers were 
permitted to transfer and to buy and sell slots for a six week period^°. After this time 
slot transfers continued, and the FAA used lotteries to allocate any additional slots 
becoming available at these airports which contained special provisions for new 
entrants. In 1984 scheduling committees were reinstated; however, the same problems 
were encountered as before. 

Given the increasing difficulties in reaching a consensus among scheduling 
committee members and the successful trial of the trading system, the "buy-sell" rule 
was implemented on 1 April 1986 at the four high-density airports. Under this rule, 
carriers are able to buy, sell, trade or lease their historic entitlements of slots used for 
domestic services in a secondary market^', where trades can be one-for-one or of 
uneven numbers of slots and accompanied by a financial payment. At all times, 
however, slots remain the property of the FAA. Slots set aside for commuter services 
cannot be bought by larger carriers, and the use of slots is subject to a "use-it-or-lose- 
it" rule . Surrendered slots and others which become available are put into a pool and 
reallocated by lottery. 25% of these are reserved for new entrants"". 



Before the buy-sell rule came into effect, the FAA made 5% of slots at 
O'Hare, Reagan National and JFK airports available for reallocation to new entrants 
and incumbents with less than eight slots at these airports in response to criticism that 



It was subsequently challenged (unsuccessfully) in the courts. 

Despite the uncertainty over how long the purchased rights would be valid, 1 94 sales took place 
during this period, and at least one firm initiated a slot brokerage program (Starkie (1992); 7). 

Off-peak slots and those used for fewer than five days/ week are allocated by the FAA. Slots 
identified by the Office of the Secretary of Transportation as required for Essential Air Services (EAS) 
are allocated directly to the carrier providing the service. 

"■ Slots not used at least 80% of the time in a nvo month period must be rehimed to the FAA for 
reallocation. 

In January 1993 the definition of new entrant was broadened to include also incumbent carriers with 
only a few slots at these airports (Starkie (1998); 113). 

Each slot is also given a "priority number", assigned by lottery, which determines its priority for 
withdrawal. Slots can be withdrawn if the number required for international ser\'ices which have been 
authorised via ASAs plus the domestic slots allocated for a particular time period exceed High Density 
Rule quotas to make a sufficient number available for international the international services. Similarly, 
slots can be withdrawn for EAS operations if not enough are available. 



grandfathering slots favoured incumbents over entrants^^. It rejected claims that a 
greater proportion of slots be made available as it was concerned that services to small 
and medium-sized communities would be affected and because incumbent carriers 
had given up slots in the past to permit new entry. It also rejected calls for similar 
withdrawals and reallocations to be held periodically in the ftiture given that new 
entrants can purchase slots and be allocated them via lottery. 

In practice, however, new entrants have experienced difficulties in obtaining 
slots which are to be used for scheduled services. This is because incumbents have 
tended to lease slots rather than selling them outright, and because both leases and 
sales have generally occurred between incumbents and carriers which would not be 
expected to compete vigorously with them^^. When sales have occurred, for example, 
they have generally been between acquired carriers and their buyers", incumbents 
and their commuter partners or to all-cargo companies. The few sales of slots have 
meant that new entrants have had few opportunities to purchase slots; the large 
amount of leasing activity among related carriers has not only meant that entrants 
have been unable to lease slots, but also that incumbents have been able to avoid 
having to surrender slots under the use-it-or-lose-it rule, and hence that there have 
been relatively few slots available in the slot pool. Not surprisingly the few slots 
which have been surrendered have been non-peak slots. The extent to which 
incumbents lease slots among themselves is likely to increase in the future given the 
increasing number of alliance agreements established among carriers. 

Unlike the lATA system or the previous system used to allocate slots at the 
high-density airports, the buy-sell rule gives new entrants the right to buy and lease 
slots regardless of whether or not they already have slots. In addition, it gives the 
most efficient new entrants an upper hand, as, in the absence of government subsidies, 
the most efficient carriers will be the most profitable ones and hence be able to bid 
more, whereas under the other systems administrative discretion determines which 
carriers will be granted slots (lATA) or all carriers are automatically granted slots 
(former system used at US high-density airports). However, on the other hand, it 
prevents new entrants from successfully completing transactions by requiring them to 
buy and lease from potential competitors. Incumbents are able to determine who slots 
are sold and leased to, and hence can control the level of competition they face^^. 
Given the relatively small number of slots in the slot pool and that these tend to be 
off-peak slots, the rule still requires new entrants to purchase the majority of the slots 



" This was done by withdrawing slots from incumbents using a reverse lottery: the FAA randomly 
selected "tickets" for each hour out of a "hat", where the number of tickets each incumbent had was 
proportional to slots held. The incumbent carrier holding the first ticket selected had first choice of 
which slots to surrender, the carrier holding the second had second choice etc. The withdrawn slots 
were allocated via lotteries in March and December 1986, where each new entrants was restricted to 
drawing a maximum of eight slots at each airpon. Any slots surrendered not subsequently demanded 
were returned to carriers in reverse order of slots surrendered. 

■* Data which distinguishes between trade in air carrier and commuter slots is only available for the 
first three years after the introduction of the buy-sell rule; however, during this period approximately 
75% of all trading in air carrier slots took place berween airlines which had some sort of cooperative 
arrangement (Starkie (1992)). 

■' The introduction of the buy-sell rule coincided with a period when the US domestic airline industry- 
was concentrating as a result of mergers and acquisitions. 

"^ It is mieresting to note that in the first tiiree years approximately 15% of leased air carrier slots were 
[eased to regional carriers which operated them using small mrboprop commuter planes, preventing 
competitors from operating them using jet aircraft (Starkie (1992)). 



it requires even though incumbents were allocated theirs free of charge. This also 
means that incumbents are initially given something of equal or near value to what 
they are after to trade with. Once again it is not necessarily the case that new entrants 
are more efficient than incumbent carriers; however, the buy-sell rule hinders the 
ability of any that are to enter. 

Permitting slot leasing allows better utilisation of slots (particularly seasonal 
slots) and hence encourages efficiency while at the same time allowing a carrier to 
avoid having to surrender slots it is not currently using. It also makes it somewhat 
easier for incumbent carriers to expand their networks in response to customer 
demand than under the lATA system, as they can use slots leased to affiliated 
carriers". However, they may only be able to expand their networks up to the point 
where they are fully utilising slots they were initially allocated and did not 
subsequently sell unless they sacrifice existing services, as they will face the same 
difficulties as new entrants in obtaining new slots. This is because all carriers always 
have to obtain new slots from other carriers. 

The buy-sell rule is relatively immune to political intervention and has greatly 
relieved the administrative burden of the FAA, as its role in slot allocation at the four 
airports has been reduced to monitoring. However, the fact that all transactions occur 
among carriers (rather than the FAA) and that a substantial proportion of these are 
among affiliated carriers clouds allocation procedure fransparency, and makes it prone 
to legal challenges. 



Section II: Alternative Slot Allocation Methods 



The analysis in Section I suggests that optimally any method of allocating 
slots should be non-discriminatory, afford all parties national freatment and be 
transparent, and hence be immune to government intervention and (private) legal 
challenges. There are three ways of allocating slots which potentially meet all of these 
criteria. 

Posted Runway-Use Prices 

One way of meeting all of these criteria may be to incorporate slot value into 
take-off and landing charges. Heathrow and Gatwick have gone someway towards 
doing this, by charging a premium for runway use at peak tunes of the day and during 
peak season . However, at most airports take-off and landing charges are the same 
throughout the year, varying only by aircraft size. If airport authorities were to price 
take-off and landing slots at levels which incorporated their average value (as well as 
the average social cost of runway use^'), excess demand for slots would be 
eliminated, and slots would be allocated to their most efficient users. 



" Most leases have tended to be for relatively short periods of time. 

These rates do not var\' across aircraft size, and hence encourage the use of larger aircraft at these 
airports. 

"■' Levine (1969) points out that currently take-off and landing charges generally reflect the marginal 
rather than the average cost of using air traffic control (ATC) services. In addition, they reflect the 



Under this system the value of take-off and landing slots would be captured by 
airport-owners"'^. Regulations which currently set an upper limit on the profit margins 
airports may make from aircraft-related business would thus need to be relaxed. 
Regulatory authorities generally impose these limits on national airports and often 
insist that they be imposed on foreign airports" to prevent them from taking 
advantage of their market power by limiting capacity and extracting monopoly rents 
from airport users. Where airports are privately-owned, however, this system may 
make airport authorities reluctant to expand airport capacity: despite the increase in 
the number of aircraft movements per day, the lower charge per movement plus the 
cost of expanding capacity may mean that profits are higher when capacity is 
constrained. 

The main problem with using posted take-off and landing charges to ration 
slot usage, however, is that this relies crucially on airport authorities' ability to 
determine market values and the average social cost of runway use at a given airport 
at each time of the day. In practice it is likely to be very difficult to perfectly price 
discriminate, given the uncertainty about the nature of market demand. Jones, Viehoff 
and Marks (1993) state that under such a system prices would be reset periodically on 
the basis of observed outcomes. For example, if there was excess demand for slots in 
a particular time interval in one period, prices would be adjusted upwards in the next 
period. It may take several periods, however, before the "correct" prices are set, such 
that scarce resources are wasted (if prices are set too high) or excess demand remains 
(if prices are set too low) in the meantime. Indeed, prices which equate demand and 
supply in every time interval may never be met, given that demand for different types 
of air transport services will grow at different rates over time. 

Lottery + After-market 

An alternative method of allocating slots which potentially meets the criteria 
set out above is to use a lottery to initially randomly allocate slots, and then permit 
post-allocation trading among lottery-participants. 

There are, however, several disadvantages to using lotteries as a method of 
slot allocation. The main disadvantage is that it is highly likely that carriers will be 
allocated non-efficient if not non-workable combinations of slots in the initial lottery. 
The more inefficient the initial allocation, the greater the extent of trading carriers will 
need to undertake in the after-market in order to obtain efficient allocations. When 
licences to operate radio spectrum were allocated by lottery in the US, for example, it 
took over two years for secondary-market trading of licenses to cease. This not only 
imposes costs on participants, but also can delay the implementation of services 

(marginal) private cost of runway use, rather than the (average) social cost, as they do not incorporate 

the costs of aircraft "footprint" pressure (runway wear-and-tear), noise or pollution. 

'" Airport-owners would capture an amount equal to the valuation of the second-highest bidder for each 

slot; successful bidders would capture the difference between their valuation and the amount they paid 

for each slot (the second-highest bid). 

'" This is done by including a clause regarding airpon pricing in agreements governing trade in air 

transport sen-ices. .Article 10.3 of the US-UK .^S.A (Bermuda 11). for example, states that airport 

charges "... may reflect, but shall not exceed, the full cost... of providing appropriate airport and air 

na\-igation facilities, and may provide for a reasonable rate of return on assets, after depreciation." 

(Jones, Viehoff and Marks (1993); 54). 



associated with the use of the slots, causing carriers to forego revenue and imposing 
huge costs on passengers and communities which rely on the provision of air transport 
services. These costs will be even larger if individual lotteries are held for each 
airport, each of which only covers the slots available at this airport, and these are 
conducted sequentially. The efficiency of the process can be maximised by imposing 
eligibility requirements (such as requiring potential lottery participants to register and 
show that they are able to operate the slots should they be allocated them) as these 
will minimise speculative behaviour''''; however, these have to be of form which does 
not deter genuine participation. 

Carriers will also encounter similar problems in the secondary market as US 
carriers currently do under the buy-sell rule, as they must obtain slots from lottery- 
winners, many of whom will be potential competitors. Carriers may tend to lease slots 
for short periods of time rather than sell them outright, and only lease them to carriers 
with whom they have established cooperative agreements. If the lottery is once-off, 
carriers may experience difficulties adjusting their slots portfolios or acquiring 
additional slots in the future, preventing them from fully responding to changes in 
consumer demand. It is thus not necessarily the case that secondary market trading 
will ever enable carriers to obtain efficient slot allocations. Trading in the secondary 
market will thus be non-transparent and subject to legal challenges. 

This process will be unpopular with some incumbent carriers, especially the 
less slots which are grand-fathered and the more are allocated by lottery, as it is likely 
that they will be forced to engage in secondary market trading in order to obtain the 
slots they require, and, if successful, they will be required to pay these slots. 
However, it may also be unpopular with the general public, as this method of slot 
allocation allows the value of a scarce public resource to be captured entirely by 
lottery winners. The government is missing out on a windfall revenue gain which 
could be used to improve the budget bottom line and bring relief to tax-payers. In the 
radio spectrum lotteries the US Government was severely criticised for giving away 
this revenue when it became public knowledge the sums of money licences were 
being bought and sold for in secondary markets. 

Auctions 

A third method of allocating slots is to use an auction process. Carriers would 
submit bids to a competition or air transport regulatory authority, and slots would be 
allocated to those with the highest bids. Carriers would thus be forced to pay market 
prices for all slots they require. Secondary market trading would be permitted to allow 
carriers to make minor adjustments to their slot portfolios in response to information 
which becomes available after the initial auction. 

The US Federal Communications Commission (FCC) has held auctions on 
sixteen occasions since July 1994 to allocate almost 6000 radio spectrum licences for 
use in nine different wireless and satellite categories. These auctions have been highly 
successful, and the FCC now has a fully computerised Automated Auction System 
(AAS) which it uses to determine the revenue-maximising configuration of bids. In 



■"' In the US radio spectrum lotteries, it took 20 months to screen potential lottery panicipants; however, 
when pre-lottery screening was abandoned the lotteries attracted approximately 400,000 speculative 
participants hoping to acquire a license to sell on. 



addition, the task of allocating radio spectrum is in many ways similar to the task of 
allocating take-off and landing slots, given their similar characteristics. 

As in the case of take-off and landing slots, radio spectrum had previously 
been allocated via administrative decisions . This was becoming increasing complex 
and time-consuming to all involved, however, as the advent of new communications 
technologies and services increased the demand for spectrum. To relieve its 
administrative burden the FCC then used lotteries. However, as already mentioned it 
took some years before after-market trading resulted in the licences being allocated to 
those even capable of and intending to providing telecommunications services, and 
the US Government was severely criticised for "throwing away" windfall revenue. 

The auction format used was a simultaneous, multi-round sealed-bid auction. 
A simultaneous format was used because spectrum licence values are to some extent 
interdependent: the value of a particular licence will depend on the value of others. 
They thus allow prices of similar items to equalise. Multiple rounds of bidding were 
conducted for two reasons. Firstly, it was thought that there was a great deal of 
uncertainty among bidders about market demand and hence the underlying value of 
the licences, that is, that common value uncertainty was high. A multi-round format 
would enable bidders to observe the bidding behaviour of their competitors, and 
hence learn more about the true value of licences in successive rounds. This, in turn, 
would reduce the risk of the "winner's curse": that the bidder winning the auction for 
a particular licence is the one who overestimates common value the most, and hence 
is not necessarily the most efficient user. Common value uncertainty also reduces 
revenues to the auctioneer, as the optimal bidding strategy for each bidder is to reduce 
their bids. Secondly, a multiple-round format enables auction participants to fully 
respond to price information obtained in later rounds. If bidding for a particular 
licence goes above the constraints of their budget, they can switch to bidding for a 
different licence. This type of auction thus reduces the risk that bidders are unable to 
respond to price information which becomes available and, given their budget 
constraints, miss out on obtaining licences entirely. Bids were required to be sealed to 
minimise the risk of collusion among auction participants; auction participants were 
assigned bidder numbers. However, after several rounds participants were generally 
able to match bidder numbers with names from bidding behaviour given a priori 
information about each bidder'*. 

Similar auction formats had been used to allocate radio spectrum in other 
countries prior to 1994. These have been conducted with varying degrees of success; 
however were they have been unsuccessful it is generally due to flaws in the design of 
the auction, rather than a failure of the actual auction mechanism itself. In New 
Zealand, for example, a series of second-price sealed-bid auctions were used to 
allocate radio spectrum in the early 1990s. The auctions successfully allocated 
spectrum to those bidders who valued it most; however, the lack of a floor price for 
bids together with thin demand and large divergences in valuations among bidders 
meant that many winners ended up having to pay only a small proportion of their 



'^ The FCC held quasi-judicial comparative hearings when there were n.vo or more applicants. 

Given this and the tact that, in its assessment, the likelihood of collusion among aiiction-parricipants 
IS low . the FCC has decided that bids will be open in future auctions (Cramton (1995)).: 



willingness to pay , which created much political controversy. Establishing reserve 
prices would have achieved the same allocation of spectrum among bidders but 
increased revenues accruing to the Government. In Australia, a first-price sealed-bid 
auction was held in 1993 to allocate two satellite television licences. Once again the 
auction successfully allocated the licences efficiently; however, achieving this took 
almost a year due to the lack of penalties for default . Implementation of the services 
associated with the licences was thus delayed by almost a year. 

Take-off and landing slots are similar to radio spectrum in a number of ways. 
Firstly, slot values are highly interdependent: the value of a particular slot to a carrier 
will depend crucially on what other slots the carrier acquires. Secondly, there is likely 
to be a great deal of uncertainty among carriers about market demand and hence 
underlying slot value. In addition, uncertainty about a particular market is likely to be 
greater among those carriers which do not currently serve that market. Furthermore, 
carriers already serving those markets will have an incentive to bid for the necessary 
slots given the investments they have made in difficult-to-transfer assets associated 
with the provision of these services. A multi-round format would reveal each bidder's 
private valuations of slots through successive rounds of bidding activity, mitigating 
uncertainty about underlying market demand and redressing the imbalance in the 
information set available to incumbents and new entrants respectively^'. It would also 
enable auction participants to fully respond to price information obtained in later 
rounds, reducing the risk that they are left with inefficient and unworkable 
combinations of slots. Sealed bidding is likely to be more necessary in slot auctions 
given the increasing number of cooperative agreements among carriers. Open bidding 
would make it easier for carriers to collude with their regional affiliates or global 
alliance partners. 

This suggests that simultaneous, multi-round, sealed-bid auctions (together 
with an after-market to permit minor adjustments) could possibly be used to allocate 
take-off and landing slots. However, there are several important ways in which the 
slot allocation problem differs from problem of how to allocate radio spectrum. 



Section III: Adapting the Spectrum Allocation Mechanism to Auction Slots 



Combinatorial Bids 

Firstly, while there is perhaps little synergy value inherent in operating 
particular combinations of spectrum licences together, this is not the case for take-off 
and landing slots. A certain combination of slots may enable a carrier to operate a 
high frequency service on a particular route, for example, such that it is able to 
capture economies of traffic density. If carriers can only bid for individual slots, they 



" In one case a firm bidding SNZ 7m ended up paying only S5000, while another bidding SI 00 000 
paid only S6. 

Two bidders with no intention of launching services put in a range of bids which ensured that they 
won the auction; they then proceeded to default on successive bids while seeking profitable resale 
opportunities. 

.An activit>' rule would need to be imposed which requires auction participants to increase their bids 
by a minimum percentage each round (which would be lowered as bidding. activity slows) in order to 
prevent carriers from concealing their true valuations until the final round. 



will have to decide how to spread this synergy value across their bids for the 
individual slots which make up such combinations. This creates the risk that a carrier 
may spread the synergy value in such a way that it is outbid on a particular slot and 
hence is left with a combination of slots which has a lower overall value than what it 
paid for them. In addition, this combination of slots may be unworkable. 

One way to avoid this problem is to permit carriers to submit combinatorial 
bids as well as bids for single slots. A combinatorial bid would be successfiil if the 
price offered for a group of slots was more than the sum of the highest bids offered 
for each of the slots individually. However, permitting combinatorial bidding may 
actually produce some inefficiencies due to free-rider problems. Two bidders each 
bidding for a single slot may have a combined valuation for the two slots which is 
higher than that of a bidder which submits a single bid for both slots, but the 
combinatorial bid may win as each of the bidders after a single slot has the incentive 
to let the other raise the bid. In addition, combinatorial auctions are difficult to 
conduct in practice, due the complexity of determining the revenue-maximising 
configuration of bids, particularly when there are many items being auctioned as 
would be the case in slot auctions. However, given that the FCC has been instructed 
to experiment with combinatorial bidding and hence that combinatorial bidding may 
be permitted in future spectrum auctions, it is possible that software capable of 
running such auctions will soon be available. 

Contingency Bids 

The second way in which the task of allocating slots differs from the problem 
of allocating radio spectrum is that slots within a particular time-period are substitutes 
for one another. The longer the length of the route a carrier plans to use particular 
slots for, the greater this time period will be. Any slot auction will thus also need to 
permit contingency bids. These will allow a carrier which wants to obtain a group of 
slots sometime within a particular time interval to submit multiple bids for this group 
which differ by time or other factors, but only be allocated at most one group of slots. 

Allowing carriers to submit contingency bids increases the complexity of 
determining the solution to the slot allocation problem. Jones, Viehoff and Marks 
(1993) also question whether carriers would be able to determine all the bids they 
could possibly submit, given that each service could potentially vary by departure 
time, aircraft size, and so on. The set of all possible bids could be so large that it is too 
time-consuming to determine; however, if all the alternatives are not considered and 
bid for, they may miss out altogether"*". 

A further problem with combinatorial auctions which permit contingency 
bidding is that no set of prices can be determined which will separate bids that are 
chosen from those that are not because this is a discrete programming problem. Only 
a lower price below which no bids are accepted and an upper price above which all 
bids are accepted can be determined. Which of the bids lying in the region between 
these tvvo prices (the "core" region) will be accepted and which rejected will be 
determined by the exact requirements of each bidding airline. Those carriers whose 
services "fit in" will be allowed to operate services at a particular time slot, even 



""* Jones, Viehoff and Marks (199?); 47 



though the bids of others whose services do not fit in may be marginally higher. In 
practice, then, some administrative assistance may be required to determine which of 
the services whose bids lie in the core region will fit in and hence be allocated the 
required slots. The decisions of these administrators will need to be transparent, non- 
discriminatory, and consistent with national treatment for the outcome to be efficient. 
However, bids lying in the core region comprise a small percentage of all bids and are 
known to decrease in relative number as the problem size increases'". 

Rassenti, Smith and Bulfm (1982) conducted controlled experiments in which 
six participants were required to determine and bid for combinations of slots available 
at six hypothetical airports. Each experiment consisted of sequence of market periods 
conducted within a three hour time limit. In each period bids submitted were entered 
into a computer which subsequently determined the revenue-maximising combination 
of bids and hence the auction solution. Those participants submitting the bids which 
maximised system revenue paid a price equal to the sum of the marginal (shadow) 
prices (the lower bound prices) of each of the slots in the combination bid for, and 
hence were able to fully capture the synergy value inherent in each slot combination. 
This charging system was adopted to encourage demand revelation: the optimal 
bidding strategy under such conditions is to bid truthfully''^. Participants could then 
trade slots in an after-market of the oral-bid type. 

Despite the potential problems associated with the existence of a core region 
of bids, the experiments were highly successfiil: experienced participants achieved 
allocative efficiencies of 98-99% of the possible surplus after only a few time periods. 
This was achieved despite the fact that inexperienced participants repeatedly tried to 
engage in speculative behaviour early on. Efficiency improved over time in each 
experiment, suggesting that learning effects were significant. Post-auction trading was 
minimal and decreased over time, despite the potential for each participant to engage 
in speculative behaviour. Indeed, such behaviour decreased over time due to the 
difficulty of obtaining the additional slots required to make-up a particular 
combination or off-loading unwanted slots in the secondary market. Furthermore, 
auction performance did not seem to deteriorate as the complexity of formulating 
combinations increased. The auction thus appears to minimise the extent to which it is 
necessary to engage in secondary market trading, which is what we would want it to 
do in practice given the difficulties inherent in trading with potential competitors and 
the cost involved in such transactions. 

Internationai Cooperation 

The third way in which the slot allocation problem differs from the task of 
allocating radio spectrum is that slots at multiple airports located in many different 
countries would need to be auctioned simultaneously. Computer software and 
communications technology, however, make this task relatively easy. Carriers and 
communities could simultaneously submit bids from all over the world via telephone 
or the Internet and monitor auction progress on the Intemet. The FCC's AAS could be 



" Rassenti, Smith and Bulfin (19S2); 406 



'" If participants were forced to pay the amount they bid in the auctions, they would ha\e tended to bid 
less than their tnie valuation of the slots. They would do this to minmiise the price they would have to 
pay should they be successful while at the same time ensuring thai the probabilin.- of success remained 
hish. 



adapted for slot allocation, significantly reducing the once-off cost associated with 
auction design. However, even if new software had to be developed, its total cost is 
likely to be equivalent to only a small proportion of the total revenue generated from 
the auctions''\ 

Decisions would need to be made as to who would fund the costs associated 
with the development of the necessary software and the running of the auctions, and 
who would actually conduct the auctions. The US may be unwilling to hand over the 
software it has developed for auctioning radio spectrum, or indeed to adapt this for 
slot auctions, without financial compensation; some sort of joint-venture-type 
agreement may need to be established, whereby several countries fund the 
development of the slot auction software. Auctions would presumably need to be run 
by a national competition authority, such as the US Department of Justice (DoJ), 
under the oversight of competition authorities of other countries. 

What is likely to be more difficult is gaining the necessary approval and 
funding to implement a slot allocation system based on auctions worldwide. Even 
implementing it initially only in the regions providing the majority of the worid's air 
transport services would require the agreement of many countries. Despite the 
benefits inherent in switching to an auction system, it is possible that incumbent 
carriers will oppose its introduction given that current ways in which slots are 
allocated protect them from significant new entry, provide them with substantial 
certainty over their future operations, and do not force them to pay for the majority of 
slots they use. Historically incumbents have been extremely effective lobbyists, given 
that they are generally large employers and given their importance to the business 
community and the travel industry, and hence their importance to the economy as a 
whole. 

New entrants may also be unsupportive of a system which forces them to pay 
for slot usage, despite the fact that it also forces incumbents to pay and requires all 
earners to deal with a neutral seller to obtain slots. Indeed, it is commonly claimed 
that new entrants will be unable to compete with larger, more established carriers in 
slot auctions, given incumbents' access to large financial reserves. This is somewhat 
misleading, however, in two ways. Firstly, in bidding for particular slots it is not just 
the absolute amount bid relative to competitors which a carrier takes into account, but 
also the difference between the profit it expects to make from operating those slots if 
it is allocated them and their cost. For an incumbent to outbid a new entrant it must 
thus not only bid a higher amount, but also still be making a profit at this level. If a 
new entrant has relatively lower costs such that its expected profit margin is greater 
than the profit margin the incumbent anticipates, it will have greater bidding leverage 
against the incumbent. Secondly, in practice, new entrants tend to be carriers 
established by successful entrepreneurs backed up by business empires and/ or 
financial institutions given the substantial costs involved in setting up an airline*''. 
They thus also generally have access to substantial funds and good credit-ratings with 
major financial institutions. 



"' The total cost of all FCC auctions to September 1997, including the costs insoh'ed in running the 
auctions, was SUS 74 million, which is equivalent to only 0.62°'o of total auction revenues raised. 
" Examples include Virgin .Mlantic Air^vays (the \'irgm Group) and Eva .Air (Evergreen shipping). 



Both incumbent and new entrant-carrier objections to the implementation of 
an auction system may decrease over time, however, given that demand for air 
transport services is predicted to continue growing at high rates and hence airport 
constraints are likely to worsen. It may also be possible to gain support for a slot 
auction system in world trade negotiations if air transport issues are negotiated 
together with issues affecting other sectors, as any concessions granted which relate to 
air transport can be balanced by benefits gained in other areas. In addition, 
particularly in the US, consumers groups wanting lower airfares and communities 
wanting more flights into particular regions regardless of the nationality of the carrier 
providing them are becoming more visible and hence are having greater influence in 
government decisions. 

Division of Auction Revenue 

The fourth way in which slot auctions would differ firom the US radio 
spectrum auctions is that auction revenue would need to be divided among many 
countries, given that the auction involves slots at airports located all over the world. 
Presumably the revenue raised from the auction of slots located at a particular airport 
would go to the government of the country in which that airport is located, given that 
governments generally hold proprietary rights over the slots available at airports 
located within their borders. If carriers are permitted to fully capture the synergy 
value inherent in particular combinations of slots as in the Rassenti, Smith and Bulfin 
(1982) experiments, governments will only receive an amount equal to the sum of the 
marginal prices of each of the slots available at the airports located within their 
borders. However, this is not likely to be controversial given that revenue 
maximisation is not the primary objective of governments in the slot auctions. 

Importantly, governments must announce what they intend to use the revenue 
accruing to them from the slot auctions for before the auctions take place, such that 
this information can be incorporated into bids. Governments may, for example, decide 
to use it to expand airport capacity. If this is the case, the present value of the future 
profits of carriers holding slots at those airports will decrease, constraining carriers' 
budgets and hence the maximum amount they will bid in the auctions. Alternatively, 
they may decide to use it to improve national accounts, and subsequently lower 
corporate tax rates. In this case carriers serving airports located in countries where this 
occurs will expect their future profits to be higher and adjust their bids upwards. 

Slot Validity 

The fifth way in which slot auctions would need to differ from the radio 
spectrum auctions is in terms of the length of time the rights allocated are valid. 
Spectrum licences are allocated for ten years; however, they are typically renewed for 
a negligible charge provided certain requirements are met and hence are virtually 
valid in perpetuity. If this was also the case for auctioned slots, even if after-market 
trading was permitted, carriers would have difficulty accessing slots after the auctions 
as slot suppliers would also be potential competitors. As already discussed, carriers 
will thus find it difficult to expand their services in line with consumer demand. 

Slot rights should therefore probably be granted for a fixed length of time 
only, and the more rapidly air transport markets are changing, the shorter this length 



of time should be. This time-period should be sufficiently long, however, to give 
auction winners the incentive to make sunk investments in the route. This must also 
be determined and announced before any auctions take place to enable auction- 
participants to incorporate this information in their bids. 



Section IV: Incorporating Route Rights in Slot Deflnition 



The final way in which slot auctions would differ from the radio spectrum 
auctions is in terms of what is being auctioned. Auctions of radio spectrum embody 
the right to provide services which require spectrum as well as the spectrum itself. 
Slot auctions, however, would only provide auction-winners with the ability to take- 
off or land at a particular airport at a particular time. The routes carriers can serve 
using the slots allocated to them are currently determined by the terms of bilateral 
agreements and the decisions of national regulatory authorities responsible for 
allocating negotiated increases in capacity. 

Under the EEA CAM agreement, carriers registered in any of the signatory 
countries have the rights to automatically begin operating new services within the 
EEA or increase capacity on routes already served. Similarly, in markets governed by 
liberalised bilateral agreements, signatory country carriers can automatically begin 
operating new services between the two countries or increase flight frequency on 
existing routes. In practice, however, whether or not they will add flights will depend 
on whether they can obtain the slots required to provide the service, or, where airports 
are severely constrained, whether they are prepared to sacrifice existing services and 
use those slots. In markets governed by ASAs, however, any negotiated capacity 
increases must be allocated among national carriers. 

In countries which have a single flag-carrier allocation is automatic. In 
countries which have multiple national carriers providing international services, 
however, regulatory authorities must decide how capacity should be allocated among 
them. Capacity has been allocated in different ways. Historically Canada had a 
"Division of the World" (DOW) policy whereby international capacity was 
automatically assigned to the national carrier which had the rights to serve that region. 
The right themselves had been pre-allocated by regulatory authorities. The lack of 
overlap of the rights granted to Air Canada and Canadian Airlines respectively, 
however, limited each carrier's exposure to competition on international routes'*^. The 
Republic of Korea uses a system based on route traffic thresholds whereby new entry 
by a national carrier (Asiana) is permitted on a route if annual traffic levels exceed a 
certain level predetermined by national regulatory authorities'*'^. Given that these 
threshold levels are known, the incumbent carrier (Korean Air) is able to limit 
competition by keeping annual traffic loads just below them. In the US the FAA holds 
quasi-judicial hearings to determine which of the competing carriers most closely 
satisfies predetermined criteria and hence should be allocated the capacity"*^. In the 
EEA national regulatory authorities allocate negotiated increases in capacity in non- 



'• For details see Oum (1995); 96-97. 
■"^ For further details see Kim (1997). 



See the US DoT Office of the Secretar>' website (' htrp:' \v\\\v.doi.govueneral orders i for transcripts 
of capacity allocation decisions. 



intra-EEA markets among their national carriers. In the UK, for example, the UK 
Civil Aviation Authority (CAA) uses quasi-judicial hearings to allocate increases in 
capacity in these markets. 

Given that it is not necessarily the case that the carriers which have been 
allocated the rights to serve particular markets will be the most efficient providers of 
services in those markets, the pool of carriers bidding for particular slots at each 
airport will not necessarily be the most efficient users of those slots. Indeed, generally 
the lower the number of carriers with rights to serve particular markets, the lower the 
probability that any slot auction outcome will be efficient. Allocating route rights 
efficiently will also not necessarily ensure that the auction outcome is optimal, as 
these routes have been pre-specified by aviation regulatory authorities. It is possible 
that rights currently available do not cover all the segment combinations carriers 
would choose to fly if current foreign investment restrictions (and hence mode of 
supply restrictions) were removed. 

One way of ensuring that service rights and hence slots are allocated 
efficiently is to incorporate the right to provide services in slot definition. Slots would 
then embody both the right to provide services and the ability to physically commence 
or terminate a service. As well as enabling carriers to determine their own segment 
combinations and the level of capacity to provide on each of these, this would turn 
what is now a two-step process into a single step. 

Consolidating these two steps would require all countries to remove the 
restrictions on capacity and foreign investment inherent in agreements they have 
concluded which govern trade in air transport services, which would essentially mean 
the end of all such agreements. Opposition to such moves is likely to be strong. 
However, the conclusion of liberalised bilateral agreements show that it is not 
impossible to remove capacity restrictions on a reciprocal basis. Similarly, the 
establishment of the EEA CAM shows that restrictions on foreign investment can be 
reciprocally removed. The conclusion of the GATS in the Uruguay Round of world 
trade negotiations also suggests that air transport services trade may not always be 
exempt fi-om the rules of the worid trading system. 

In theory, such a system would open-up bidding for all slots to all carriers. 
However, in practice the number of carriers bidding for particular slots will be 
determined by passenger demand and the costs of providing services in particular 
markets. While it may be possible for a particular carrier to provide services between 
two particular cities, they may not bid for the slots which would enable them to 
operate such services as they would be unable to make a profit on these. This may be 
because demand for such services is low, or because the service does not connect with 
their existing network; studies have shown that economies of scale are negligible or 
even negative in the provision of air transport services in the sense that adding a non- 
contiguous route to a carrier's network does not reduce its average costs. It is thus 
highly likely that, at least initially, the number of carriers bidding for slots at each 
airport will not be much higher than the number which currently apply for slots at 
these airports under the lATA system or the buy-sell rule. In the longer-term these 
numbers may increase as carriers' networks expand in line with traffic increases and 
the financial resources available to each increases. 



Summary and Conclusions 

In summary, it may be useful to re-examine the feasibility of using auctions to 
allocate take-off and landing slots at the world's major airports in light of the highly 
successful auctions of radio spectrum in the US. Given the similarities between the 
tasks of allocating spectrum and the slot allocation problem, it is possible that we are 
not too far away from developing the software required to handle slot auctions. The 
results of the controlled experiments reported in Rassenti, Smith and Bulfin (1982) 
show that it is possible to achieve extremely high levels of efficiency in slot auctions 
which permit combinatorial and contingency bidding despite the potential problems 
associated with the existence of the core region. Such auctions appear to require 
secondary market trading only to correct (marginal) misallocations and to permit 
adjustments in response to information not available at the time of the auctions, and 
hence avoid the problems generally associated with secondary markets such as being 
forced to try to obtain slots from potential competitors. However, further 
experimentation with combinatorial bidding will be required to determine the effects 
of the free-rider problem on auction efficiency. 

Perhaps the biggest hurdle to moving to auction-based slot allocation system 
and ensuring that its outcomes are fiilly efficient will be obtaining the approval of the 
many countries in which the world's major airports are located. Their approval will be 
required to remove the lATA system and the buy-sell rule, as well as the capacity and 
foreign investment restrictions contained in the agreements which currently govern 
trade in air transport services. Multilateral cooperation will also be required to 
develop the necessary software and to run and monitor the auctions. Agreement may 
be more forthcoming over time, however, as airports become increasingly 
constrained, and as more of air transport services trade becomes governed by 
agreements with liberal capacity and foreign investment provisions. Increasing 
jurisdiction of the GATS over world services trade will also put pressure on countries 
to agree. 

Whether or not auction outcomes are efficient in practice will also depend on 
the extent to which carriers have competitive access to other facilities essential to the 
provision of air transport services, such as airport infrastructure and services and 
ticket sales channels. If particular carriers anticipate problems accessing these 
facilities at competitive prices in relation to providing services to particular 
destinations, they will scale down their bids for slots at the associated airports. It will 
thus be useful in future studies to examine ways in which the problem of accessing 
these facilities on competitive terms can be addressed. It may also be useful to 
examine whether there are complementarities between slots and some of these 
facilities, such as terminal gates and ground-handling services, in order to determine 
whether or not slots should be auctioned jointly with other facilities. 



Bibliography 



Balinski, M and F Sand (1 985), "Auctioning Landing Rights at Congested 
Airports", in Young, H (ed) Cost Allocation: Methods. Principles. Applications. 
Elsevier Science Publishers (North Holland); 179-192 

Caves, D, L Christensen and M Tretheway (1984), "Economies of Density 
Versus Economies of Scale: Why Trunk and Local Airline Costs Differ", Rand 
Journal of Economics. 15 (4); 471-489 

Cramton, P (1995), "Money Out of Thin Air: The Nationwide Narrowband 
PCS Auction", Journal of Economics and Management Strategy, 4 (2), Summer- 267- 
343 

Gale, I (1994), "Competition for Scarce Inputs: The Case of Airport Takeoff 
and Landing Slots", Federal Reserve Bank of Cleveland Economic Review 30 (2)- 18- 
25 

GRA Inc (1996), Market Effects of the Proposed American Airlines/ British 
Airways Alliance, July 

Grether, D, M Isaac and C Plott (1989), The Allocation of Scarce Resources: 
Experimental Economics and the Problem of Allocating Airport Slots, Underground 
Classics in Economics Series, Westview Press 

International Air Transport Association (lATA) (1998), Scheduling 
Procedures Guide. Twenty-Third Edition, January 

Jones, I, I Viehoff and P Marks (1993), "The Economics of Airport Slots" 
Fiscal Studies, 14 (4); 37-57 

Kim, J (1 997), "Multiple Designation Policy in Korea", in Chia, Findlay and 
Singh (eds), Asia Pacific Air Transport, Institute for South East Asian Studies 
(ISEAS), Singapore 

Levine, M (1969), "Landing Fees and the Airport Congestion Problem", 
Journal of Law and Economics, 12; 79-108 

McMillan, J (1994), "Selling Spectrum Rights", Journal of Economic 
Perspectives. 8 (3), Summer; 145-162 

Milgrom, P (1998), "Game Theory and the Spectrum Auctions", European 
Economic Review. 42 (3-5), May; 771-778 

Morrison, S and C Winston (1995), The Evolution of the Airline Industry. 
Brookings Institution, Washington DC 



Oum, T (1995), "An Analysis of the Division of the World Policy for the 
Canadian Airline Industry", in Oum, T (ed) Airline Economics and Policy: Selected 
Papers o/Tae Hoon Oum, Transportation Series 105, Korea Research Foundation for 
the 21" Century 

Rassenti, S, V Smith and R Bulfin (1982), "A Combinatorial Auction 
Mechanism for Airport Time Slot Allocation", The BellJournal of Economics, 13 (2); 
402-417 

Starkie, D (1998), "Allocating Airport Slots: A Role for the Market?", Journal 
of Air Transport Management, 4; 111-116 

Starkie, D (1992), Slot Trading at United States Airports, Putnam, Hayes and 
Bartlett Ltd Economic and Management Counsel 

Starkie, D (1994), "The US Market in Airport Slots", Journal of Transport 
Economics and Policy, September; 325-329 

Tretheway, M and T Oum (1992), Airline Economics: Foundations For 
Strategy and Policy, Centre for Transportation Studies, University of British 
Columbia, Vancouver 

UK Civil Aviation Authority (CAA) (1998), The Single European Aviation 
Market: The First Five Years, CAP 685, London, June 

Warren, T, V Tamms and C Findlay (1999), "Beyond the Bilateral System: 
Competition Policy and Trade in International Air Transport Services", paper 
presented at the American Economic Association Annual Meeting and Meeting of the 
Transportation and Public Utilities Group, New York, 3 January 



Regulation as a driver for international airline alliances 
Hannu Seristo 

ABSTRACT 

There is an apparent strong need to consolidate the airUne industry of the world. However, 
there have been few mergers and acquisitions in the industry; particularly international 
mergers and acquisitions are almost non-existent. Airlines have used other ways than 
mergers and acquisitions, primarily alliances, in their search for more competitiveness. 
Evidently much due to government control airlines can go only so far in trying to rationalise 
and consolidate the industry and they appear rather frustrated by the fact that the natural 
evolution towards transnational companies in the industry is effectively blocked. 

So, the airline industry is a major theatre for alliances. Many reasons for alliances have been 
suggested, such as the seek for more market presence and lower costs; the role of authority 
intervention in the industry restructuring has been often noted, too. However, it appears that 
there are no clear frameworks for assessing alliances in the framework of airline strategies - 
covering the drivers or motives and objectives of alliances. Consequently, this study will 
focus on the international alliances in the airline industry, and sets out to seek answers to the 
following questions: 

• what are the drivers of international airline alliances? 

• what are the objectives of international alliances? 

• what is the role of government or other authority regulation in the alliance formation 
within the airline industry? 

The objective of this study is to present a framework depicting drivers and key variables of 
international airline alliances, particularly the authority regulation, in the framework of 
airline strategies. The core of the study is a longitudinal analysis of the alliances reported in 
the industry, using firms' own and third party material as sources of information. Key 
sources of information are annual reports between 1988 to 1998 from thirteen major airlines 
from Europe, North America and Asia. 

As to the findings, among the three key alliance motivators it appears that it is the pursuit of 
stronger market presence that clearly has been more apparent and dominant, the need for 
better resource utilisation being clearly secondary and also more of a longer-term nature. 
Concerning the role of regulation and the need to circumvent it, the assessment is somewhat 
difficult. It seems very often to be a fundamental factor, but very rarely is it expressed as the 
primary reason - it could be seen a relevant, compelling factor for nearly all airlines of the 
world, as alliances appear the only feasible way to grow and seek presence in a larger market. 
There are differences between airlines from North America and Europe, as well as between 
large and small carriers, concerning whether they see alliances primarily as offensive or 
defensive moves. 

The paper presents a framework where drivers, or motives, and objectives for airline 
alliances are presented in a corporate strategy setting; special emphasis is on the role of 
regulation. Also airlines of different size are positioned along the offensive-defensive 
dimension of alliance objectives. 



Key %vor(l.s: international airline ulliances. motives and obiectives. reculation. 



Regulation as a driver for international airline alliances 

Hannu Seristo 



INTRODUCTION 

The growth of alliances in the 1990's has been rather phenomenal; studies suggest 
annual growth rates above 100 per cent in the number of business alliances (see e.g. 
Pekar & Allio 1994, Luo 1996). Strategic alliances have been found to be unstable 
and they generally speaking have a poor record of success (see e.g. Gant 1995, 
Brouthers et al. 1995). Strategic alliances have been studied from various perspectives 
- e.g. that of alliances characteristics (Borys and Jemison 1989), complexity of 
alliances (Killing 1988), rationale of alliances (Contractor and Lorange 1988), 
transaction costs (Parkhe 1993), alliances between competitors (Hamel, Doz and 
Prahalad 1989), trust and contractual arrangements in alliances (Gulati 1995), learning 
from alliances (Parkhe 1991; Lei et al. 1997; Inkpen 1998), value creation through 
alliances (Chan et al. 1997; Doz and Hamel 1998), and the assessment of aUiance 
performance (Dussauge and Garrette 1995; Gleister and Buckley 1998). 

The airline industry is a major theatre for alliances. Airline alliances have been 
studied, for example, from the perspective of benefits (see Park and Zhang 1997), 
performance enhancement (Park and Cho 1997), corporate value (Park and Zhang 
1998), critical success factors (Bissessur 1996) and safety implications (Button 1997). 
Many reasons for alliances have been suggested, such as the seek for more market 
presence and lower costs. Also the role of authority intervention in the industry 
restructuring has been often noted as one contributing factor to the popularity of 
alliances recently. However, it appears that there are no clear frameworks for 
assessing the relationship between airlines' strategies and the alliances - including the 
drivers or motives and objectives of alliances. Consequently, this study will focus on 
the international alliances in the airline industry, and sets out to seek answers to the 
following questions: 

• what are the drivers of intemational airline alliances? 

• what are the objectives of intemational alliances? 

• what is the role of government or other authority regulation in the alliance 
formation within the airline industry? 

A brief history of airline alliances 

The world airline industry has seen very strong growth during the last decades. For 
example, the volume of scheduled services, measured in the number of passenger- 
kilometres flown, more than doubled from 1980 to 1995. As for the future, most 
forecasters see an annual growth rate of traffic volume in the region of 4 - 5 per cent, 
meaning that the traffic volumes would again double in about 15 years. 

The airline industry has experienced major changes in the operational environment 
during the last two decades. Liberalisation, or deregulation, has changed the rules of 
competition drastically in most major markets of the worid. The industry experienced 
a severe recession in the early 1990's, sparked by the Persian Gulf crisis, but with the 
recovery of major economies and the ver>' strong growth in air transport demand it has 



improved performance significantly towards the end of the decade. In fact, in 1997 the 
100 largest airlines of the world had a combined sales of USD 288 billion, operating 
profit of 18.6 billion, and net profit of 9.5 billion. In 1992, the worst financial year of 
the industry history, the corresponding net result was a loss of USD 8 billion. Still, the 
financial performance leaves room for improvement - the net margin for the top- 100 
carriers in 1997 was only 3.3 per cent (Gallacher 1998). 

A great majority of the existing alliances in the airiine industry have been formed in 
the 1990's, but there are alliances the origins of which can be traced as far back as to 
the 1940's. For example Air France has helped to set up the operations of many 
African airlines - such as Air Afrique, Royal Air Maroc and Tunisair - and still have 
equity stakes in those carriers. Similarly, Iberia invested already in 1948 in Aviaco in 
South America. National interests and governments played a key role in these early 
alliances. There was quite little alliance activity until the late-1980's when a number 
of equity-based arrangements took place. It was the Scandinavian SAS which really 
started to proactively seek alliances, perhaps with a more strategy-level approach than 
what had been done until then by other airlines. SAS worked on many equity-based 
schemes, and had some success but some failures, too. 

In the 1990's the number of alliances has steadily grown each year, and the scene has 
become very unstable. For the sake of comparison, in 1990 the industry sources listed 
172 alliances, out of which 82 involved equity investment (Airline Business 1990). 
The latest survey by Air Transport Intelligence (1998) reported that there were a total 
of 502 airline alliances in mid-1998, with an increase of 38 per cent over the year 
1997 - these alliances were formed among 196 airlines. Most airline alliances are 
between two partners, but recently arrangements of more than two participants have 
emerged. World airlines are in the process of forming groups in their preparation for 
harder global competition - the largest groups Star Alliance and oneworld now have 
each about 20 per cent of the worid intemational passenger markets. Most alliances 
are between airlines from different countries, but there are alliances between carriers 
of the same nationality, too. Most airlines have several alliances, including domestic 
and intemational alliances - the largest number of alliances in 1998 was by Air France 
with 28 arrangements, out of which all but one were with foreign partners. 

Out of the total of 502 alliances in 1998 only 56 (11%) involved equity; government 
authorities play a key role in determining the conditions for equity-based 
arrangements. The role of government or other authority regulation, or other type of 
intervention, in the airline partnerships deserves a closer look. 

On the nature of airline alliances 

There have been very few mergers and acquisitions in the world airline industry as a 
whole. There was quite a lot of M&A activity in the United Sates in the 1980's, but 
overall, particularly intemational mergers and acquisitions are almost non-existent. 
The reason for this is the prohibitive stand by regulatory authorities world-wide. 
Consequently airlines ha\'e been forced to use other ways than mergers and 
acquisitions in their search for more competitiveness. There is a strong need to 
consolidate the industiy, but evidently much due to government control airlines can 
go only so far in tr>'ing to rationalise and consolidate the industry. 



RJioades and Lush (1997) have suggested two dimensions on which aUiance 
arrangements differ, namely commitment of resources and complexity of 
arrangement. Partly following that division, alliances can be put into three categories 
based on the extent of co-operation: simple operative route-based alliances, broader 
marketing alliances, and equity-based alliances. There are various reasons why an 
airline forms alliances with other airlines. Usually these reasons are linked to the 
strive for more competitiveness on the global market, for which airlines have used a 
number of different co-operative arrangements. Joint ground handling, co-ordination 
of schedules, joint flight operations, swap of flying personnel, sales and purchases of 
block space on aircraft, code-sharing and equity investments in another airline are 
some of the co-operative ways used. 

As mentioned, the airline industry has been very tightly regulated for most of its 
history. Liberalisation really started only in 1978 with the US airline industry 
deregulation. Even after the European Union reached the final stage of its airline 
industry liberalisation in 1997 there are many types of regulations and limitations that 
government authorities set on airline operations and competition. In fact the situation 
has reached a somewhat schizophrenic point: on the one hand authorities press for 
more competition through less regulation, but on the other hand, when stronger 
airlines try to rationalise operations in the name of better competitiveness, then 
authorities intervene and set limits on or even deny such efforts. It seems that fair play 
is sought, but not too fair. Consequently it is the authorities, primarily those of the 
United States and the European Union, that may decide whether the airline industry 
can develop into one of efficient global players, global quality service and, perhaps, 
low fares for consumers. 

Challenges and open questions 

Airline alliances, just like strategic alliances in most other industries, have had a 
rather poor record of success. It has been suggested that fewer than 30 per cent of 
international alliances in the airline industry have been successful (Lindqvist 1996). 

Simply put, airlines seek international competitiveness through alliances. Even if in 
general it is the economies - be that of scale, scope or density - that motivate airlines, 
it is not necessarily completely clear what are the different types of drivers that are 
behind alliance formation. Also, the consequent objectives of alliances are not always 
clear. Moreover, the particular role of authority regulation in alliance formation is 
often unclear. 

As suggested by earlier research, there are drivers of different level for international 
airline alliances. Some alliances are driven by mere cost savings in operations, like 
through rationalising ground handling at airports operated by both or all partners. 
Others are more of a market power issue, for example through code-sharing and 
pooled frequent flyer programs. Yet others may be more of a strategic nature, aiming 
for e.xample at the mere survival of the airline. Concerning the objectives of 
international airline alliances, the immediate objectives can naturalK' be drawn firom 
the drivers; so, the objective of a block seat arrangement with another airline would be 
to secure or increase sales. However, the longer-term strategic level objectives of, say, 



growth, market expansion, image enhancement, learning, and so on are seldom crystal 
clear. 

The objective of this study is to suggest a model which depicts drivers and key 
variables of international airline alliances, particulariy the authority regulation, in the 
framework of airline strategies. The study bases on prior research on competition in 
the airline industry and on strategic alliances. The core of the study is a longitudinal 
analysis of the alliances reported in the industry, using firms' own and third party 
material as sources of information. Key sources of information are annual reports 
between 1988 to 1998 from the following airiines: Air Canada, American Airiines, 
British Airways, Canadian Airiines (PWA Corp.), Delta Air Lines, Finnair, KLM,' 
Lufthansa, Qantas Airways, SAS, Swissair, Thai Airways International, and United 
Airlines. 



SPECIAL FEATURES OF THE AIRLINE INDUSTRY 

The airline industry is in many ways one of the most international of service 
industries. International traffic forms a major portion of all air traffic, and even 
domestic traffic is most often dependent on or at least tightly linked to international 
services. Then, do multinational enterprises dominate in this very much international 
business, like they do in most other industries? 

In this very much intemational business there are no dominant, truly global players. A 
question has often been asked whether there are true multinational companies (MNC) 
among airlines. By definition MNCs should conform to the criteria (see Bartlett and 
Ghoshal, 1995) of, first, having substantial direct investments in foreign countries. 
Second, they should be engaged in the active management of the offshore assets, 
rather than simply hold them in a passive portfolio. The management criteria would 
appear to be fulfilled by most intemationally operating carriers. However, concerning 
investments, airlines rarely have significant tangible investment in foreign countries, 
but typically hold only rather small marketing subsidiaries abroad. In addition some 
larger carriers own partly or wholly smaller carriers that operate in foreign countries. 
The investment criteria comes into an interesting light when one considers that the 
key production machinery, the aircraft, are in fact assets that move at the speed of 
some 800 kilometres per hour from one country to another - so even if each aircraft is 
always registered in a certain country the determination of where the production 
machinery really is located may be somewhat indefinite. However, as Bartlett and 
Ghoshal (1995) define the investments not only as production facilities but also as 
financial, legal and contractual relationships with foreign affiliates - in addition they 
emphasise the management integration of operations in different countries as the key 
differentiating characteristic of an MNC - it is fair to say that many intemationally 
operating large airiines are MNCs. Then, whether airiines are transnational companies 
can be questioned, too. The term transnational has often been used rather loosely, but 
the more specific definition of the term used by Bartlett and Ghoshal refers to firms 
being locally responsive in various national markets while retaining their global 
efficiency. This definition suggests specialised but dispersed resources and activities, 
realised in the form of interdependent network of world-wide operation, producing 
both efficiency and flexibility at the same time. Now, whether airlines operate as 



suggested by the transnational criteria is rather difficult to determine. Perhaps the best 
answer today is that some do, most do not. 

In air transports there still is much nationalistic thinking shown through protection by 
legislators and bargaining by unions; however, consumers of today are really global in 
their attitudes. Consumers are renting cars from Hertz because it offers convenient 
service, reliability and value for money - not primarily because the company 
originates from a certain country. Even more clearly, in the manufactured goods 
sector consumers are buying, for instance, Nokia mobile phones because they offer 
versatile features, good quality and have aesthetically appealing design - not because 
they are manufactured mostly in Finland. 

Management in any industry resist the loss of control. It is tme that in airline business, 
just as in many other service businesses, the control of, say, quality aspects of the 
product is essential. However, airlines have a long time ago given up much of the 
control in one of the key functions as the sales function is outsourced to a high degree 
in most airlines. It is true that computerised reservation systems are used by airlines to 
control the sales and airlines actively try to have an impact on the sales managed by 
travel a.eents. 



■*o^ 



The role of governments and unions 

International air traffic has been strongly affected by bilateral agreements between 
governments. As suggested by many observers, the process towards more liberal 
bilaterals, extensive multilateral arrangements, and open-skies agreements is still far 
from completed. It has been suggested that bilateral agreements have been a barrier to 
organic international growth for many airlines, and they could be considered tools of 
protectionism by nation states. The bilateral agreements have been an important 
variable when foreign ownership of airlines has been discussed; the system has 
historically built on the assumption that an airline based in and operating mainly from 
one country is also owned by parties of that country. Hence, if a British firm owned a 
US based airline wishing to operate on the Atlantic market between the US and the 
UK, the interpretation of the spirit of the bilateral agreement would be complicated. In 
fact the US government has limited the share of ownership by foreign parties in US 
airlines to 25 per cent. 

A complication in the open-skies agreements is the issue of who are the parties to the 
agreements in the case of Europe. Namely, the European Commission sees that it 
should be the signing partner in the US-Europe open-skies deal. However, many of 
the member countries of the European Union would definitely like to have agreements 
between the nation states instead of between the Union and the USA. 

In general temis, authorities have eased their regulation of the airline industry since 
the late 1970's, but they still play an important role. The liberalisation, or 
deregulation, of the industr>' has aimed at bringing competition to the market place, 
compelling airlines to better efficiency, and bringing benefits to consumers in the 
form of better offering of senices at a lower price. The same rationale is seen in 
many other industries. There, e\'entual consolidation of the industn,' has often 
followed, mosily through mergers and acquisitions. In the airline industry, however, 



authorities have been very strict about allowing particularly transnational mergers or 
acquisitions, something that puts airline management in a rather perplexing position. 
It seems that governments are very careful and protective of their national airlines - no 
matter if the government has an equity stake in an airline or not. To set this against a 
broader picture, one could ask why airlines should be treated by authorities with such 
a nationalistic ethos when the consumers, the flying passengers, no longer put much 
emphasis on the nationality of a carrier. 

The well-known argument from the authorities - and a very much understandable and 
valid one - is that the consolidation of the airline industry would spell dangers 
particularly to the consumers: division of markets between few very large carriers, 
less competition, less choice, higher fares and poorer service. In other words exactly 
those problems that were tackled through the deregulation process started in the US in 
the late 1970s. While the argumentation by the authorities makes sense, it could lead 
to a re-regulated airline industry, where the competitive pressure would not drive 
airlines to rationalised operations and efficiency. The issue is about finding the fine 
balance between allowing airlines to rationalise operations through industry 
restructuring and, on the other hand, ensuring that there is sufficient competition 
between the large airline groups in most markets. Looking back at the history of 
industry deregulation it appears inevitable that there will always be certain markets 
where competition does not work and bring the benefits to the consumer. 

It has been sometimes expressed by national authorities that the reason for being strict 
about allowing transnational mergers or acquisitions in the airline industry is the issue 
of national security. That may well be justified in the less and less common case of 
government-owned carriers, but in the case of at least partially privatised, let alone 
fully private airlines, the argument becomes somewhat old-fashioned. As noted by 
some observers, when even defence-related manufacturing industries are allowed to 
consolidate internationally, it seems strange to be so careful about air transports. After 
all, a clear distinction should be made between public services and business - military 
air lift capacity and commercial airlines should not be confused. A perspective of a 
private - and often foreign - share owner of an airline may be that he or she could not 
care less how some nation state organises her military air lift capacity. Moreover, in 
case air transport capacity needs arise for military purposes, that capacity is very 
likely available from the market, at market prices. 

Another interest party to the issue of international airiine industry consolidation is the 
labour unions. Interestingly it is the pilot unions that are working on transnational 
labour movement to protect the interest of their members. It appears that unions have 
opposed mergers and acquisitions - and appear to oppose large-scale alliances, too - 
because the M&A's and alliances might lead to more efficient organisation and thus 
lower demand for personnel. However, as suggested earlier by industry observers, it 
appears that in the light of the forecasted growth of demand for air transport it is an 
unwarranted fear that jobs would be lost in the industry as a whole. Perhaps it would 
be more con-ect to speak of only more slowly increasing need for personnel. 

.Another reason for union opposition ma\' be the teared pressure on remuneration, 
which in the airline industry is ver>' good across the board, compared to any other 



industry. The remuneration pressure would partly be due to better exposure of 
performance and efficiency; this would very likely lead to unwanted changes in the 
organisation of inefficient carriers. However, it would be in the interest of the industry 
as a whole and the consumer in particular to have changes driven through in 
inefficient carriers. 



DRIVERS OF INTERNATIONAL ALLIANCES 

In general terms earlier research has categorised the reasons for alliances, for example, 
as: 

• risk sharing 

• scale economies 

• access to markets 

• access to technology 

• market convergence. 

As to airline alliances, it has been suggested (see e.g. Alamdari and Morrell 1997) that 
there are two main drivers: first, search for more market power, and secondly, search 
for lower operating costs. These broad categories cover the two basic reasons for 
airlines' alliance arrangements, but it appears that there is room to elaborate further. 
More of an industry-level research (see Antoniou 1998) has suggested that the 
formation of mega-carriers - either through mergers or alliances - would appear to be 
a solution to the empty-core problem of the airline industry, pre-empting complete 
deregulation. 

Studying the airiine industry alliances of the past suggests that the role of authority 
control is an area which deserves to be brought up as a reason for alliances. The 
following factors - suggested in earlier research - can be seen as drivers of alliances in 
the industry: 

• mergers and acquisitions in general have been tightly controlled in the airline 
industry, 

• foreign ownership in airlines has been restricted by governments, 

• bilateral agreements between countries make foreign ownership of airlines 
problematic, 

• bilateral agreements have been a way of protecting markets. 

As to restrictions on foreign ownership in airlines, it appears to be the control that 
worries authorities. The control by a foreign investor in an airline has been restricted 
by other means than ownership limitations, too, such as the number of members to be 
appointed in the board of the airline. 

Considering their need for reaching the assumed economies of scale, scope or density, 
airlines have been left with very few options other than slow organic growth or 
alliances. In pursuing alliances airlines have due to government control had limited 
options, leading mostly to code-sharing an^angemeiits. 



Objectives of alliances 

Earlier research on strategic alliances - typically dealing with manufacturing 
industries - has pointed to two primary categories of objectives in alliance 
arrangements: product objectives and knowledge objectives. In the area of product 
objectives there appears to have been two primary goals: either the enliancement of 
product offering or the reduction of production costs. As to knowledge objectives, the 
goal has typically been to learn some specific new technology or process from a 
partner; it appears that the goals as to knowledge transfer have often been rather 
specified and particular. 



Regulation as a driver for alliances 

This study examined the alliance history of the major airlines of the worid using 
several sources of information such as general news services, industry press, and 
airlines' own publications. In order to illustrate the role of government regulation in 
the alliance development very brief summaries from a few key airiines are presented; 
the summaries are hmited to international alliance activities from 1988 to 1998. The 
summaries have quotations from only one type of information source - airiine annual 
reports to enhance the comparability of company views; namely the annual report is a 
key media to deliver messages to the investors, authorities and the public, and the 
messages need to be informative and truthful. Of course a lot is left unsaid in annual 
reports - for instance comments concerning some particular competitors - but that is 
likely to apply to any contemporary published material. 

SAS started in the 1980's with a strategy of building alliances with airiines that were 
either smaller or of the same size as SAS itself In the 1990's, under new leadership, 
the alliance strategy was refocused on partnering with large airiines. Also, SAS 
experience had taught that equity-based arrangements are very difficult to manage, 
and therefore further alliances would be pursued without ownership. Moreover, the 
emphasis in own operations was pulled back fi-om global reach to concentrate on 
being a dominant player in the markets of the home region. It appears that the Star 
Alliance is seen to produce significant benefits in the short and medium-term in the 
area of marketing, and in the longer term additional benefits are expected from 
operational cost savings in maintenance, sourcing, handling, and so on. The role of 
regulation is seldom brought up by SAS management; however, the CEO Carizon 
wrote in the 1992 annual report: 

//; the future companies which obstinately uphold national interests and 
allow them to stand in the way of essential restructuring will have chosen the 
route towards elimination. 

In its 1989 annual report AZM emphasised the liberalisation of the European aviation 
as a motive to strive for co-operative links with other carriers. These links were seen 
to provide additional opportunities in both passenger and cargo markets to safeguard 
KLM's market position. The criteria for the co-operative links comprised e.g. securing 
or expanding the position of Schiphol (Amsterdam) airport as the gateway to Europe. 
Overall KLM's approach to alliances appears to represent a rather common way of 
seeing alliances. First of all, air traffic politics and the regulation b\- government 
authorities are both a major motivator and a limitation to the pursuit of alliances. 



Secondly, it appears that the relative significance of market presence and more 
efficient resource utilisation varies by the economic turns: in good times market 
expansion appears as a key driver, but in harder times the need to reduce costs is 
emphasised. 

It appears that Lufthansa management was not very keen on tying partnership knots 
with other airlines until the recession of the early 1990's really hit the company. The 
sentiment in the company in the late 1980's and early 1990's was that bilateral air 
traffic agreements were not providing fair playing field for the world's airlines. The 
significance of regulation as a motivator for alliances by Lufthansa is apparent; on the 
other hand, regulation is seen as a hindrance to alliance building. Overall Lufthansa's 
participation in alliances appears to be justified through a combination of market 
presence and resource utilisation factors. It would appear that Lufthansa has seen 
more value than other airlines in the cost reduction possibilities that alliances may 
offer. 

American Airlines has built its international operations quite slowly, operating first 
and foremost within the USA. The strong areas for American outside the USA have 
traditionally been the Caribbean and Central and South America. Even in 1990 
American operated only to eight countries in Europe, but in the 1990's the 
international expansion has been significant. The international growth has been 
primarily internal, although American has acquired intemational routes from other 
airlines such as Eastern and TWA. American has been perhaps the most active airline 
to participate in the intemational air politics debate and has demanded more 
opportunities to operate internationally on a competitive basis, as illustrated by the 
following quotes from the 1990 annual report: 

Unfortunately U.S. airlines seeking to spread their wings in international 
marketplace face some daunting barriers, most of which - since they are 
rooted in the protectionist policies of foreign governments - can be overcome 
only by an active partnership between industry and government.... Since they 
[bilateral agreements] assign a higher prority to the welfare of national 
airlines than to the health of national economies, they are entirely 
inconsistent with today's economic realities. 
The full realisation of the alliance with British Airways has been delayed for years. In 
1997 American CEO Crandall wrote in the annual report: 

Because the airline industry is increasingly global, remaining competitive 
requires us to serve the largest possible number of origin-destination 
markets world-wide.. .The American-British Airways alliance is the 
centrepiece of a pattern of alliances we have been building as we adjust to 
the clianging nature of international competitio)i. 
The 1998 American annual report wrote: 

By granting antitrust immunity to alliances between U.S. and foreign 
carriers, the U.S. has made international alliances a virtual necessity. 
American has reacted to the changing environment by setting out to create 
the indiistiy 's premier set of alliances. 

DeltLi Air Lines purchased nearly all of the collapsed Pan Am's transatlantic routes, 
shifting its focus from being a predominantl)' U.S. domestic carrier to that of being a 



global airline. In the early 1990's Delta stated that a part of its international strategy 
was to use code-sharing with other quality airlines to support Delta's international 
service. The reasoning was that this enabled Delta to remain in markets that would be 
unprofitable to fly alone, and to offer service in new markets without major capital 
expenditures. In the annual report 1994 it was reported: 

Delta will continue to advocate a more open, market oriented operating 
environment. ..Delta's goal is to serve its customers while increasing 
efficiency and expanding market presence by developing a network of 
mutually beneficial code-sharing alliances. 
By 1995 the international code-sharing arrangements had been made with 
Aeromexico, All Nippon Airways, Austrian Airlines, Korean Air, Sabena, Singapore 
Airlines, Swissair and Virgin Atlantic Airways. In 1996 there were 13 code-sharing 
partners; with three of them Delta received approval of antitrust immunity from the 
U.S. Department of Transportation to pursue a global marketing alliance. This 
marketing alliance called Atlantic Excellence between Delta, Austrian Airlines, 
Sabena and Swissair included - in addition to code-sharing - pricing, scheduling and 
other operational co-ordination; joint sales and marketing were still seen as 
"opportunities". The 1996 annual report reported: 

...The alliance agreements establish a legal framework... to allow the four 
carriers to form a seamless transatlantic air transport system while retaining 
their unique corporate and national identities... 

In 1997 Delta announced code-sharing arrangements with Air France, China Southern 
and Transbrasil. Considering Asian operations the 1997 annual report notes: 

Delta continues to pursue additional authorities to serve Japan, but is 
impeded by the highly restrictive aviation agreement between the U.S. and 
Japan. Delta 's limited Japan sei-vices will be supplemented by additional 
service to Asia through code-sharing arrangements with China Southern and 
Korean Air. 
In 1998 Delta and United Airlines agreed on a broad marketing relationship; however, 
due to opposition from the pilots' union Delta was not able to proceed with a code- 
sharing arrangement with United, but had to continue the co-operation through 
reciprocal frequent flyer program only. In the annual report for 1998 Delta underlined 
the role of strategic alliances: 

Delta will proceed aggressively with world-wide alliance discussions in the 

future, not just because alliances are desirable from a business standpoint - 

although they are - but also because we must. Airline alliances are 

revolutionising the nature of world-wide competition, and Delta intends to be 

a leader as these changes occur. 

In about seven years Delta has grown from a domestic carrier to a significant global 

player, much thanks to alliance arrangements particularly in Europe. For example. 

Delta has been since 1997 the largest operator on the North Atlantic market, the 

largest and most competitive international market in the world. In Delta's history of 

international alliances the evasion of regulation and the market power as drivers 

appear to stand out, and the resource utilisation improvement seems to be primarily 

Delta's internal effort. 



Model for international airline alliance dynamics 

Based on the study within the airline industry a model (Figure 1) is suggested on the 
dynamics of international alliances, depicting factors that have been the drivers of the 
alliance efforts. The model is set in a framework where the relevant recent changes in 
the airline industry are shown, as well as the consequent basic strategic choices and 
the alternative strategies for airlines; the model builds on a strategy framework 
presented earlier (Seristo 1993). 

(Insert Figure 1 about here) 

There are numerous factors, many of which have been touched upon in this article, 
that effectively limit the basic strategy choices of airlines into three: growth strategy, 
focus strategy and lowest cost strategy. Growth can be sought either internally 
(organic growth) or externally. As internal growth is often slow, it may be preferable 
under the present circumstances for many airlines to seek growth externally; then the 
options are mergers and acquisitions or alliances. As there are many limitations to 
airline mergers and acquisitions, alliances provide often a less complicated route for 
growth. Alliances provide more flexibility than outright mergers and acquisition, and 
they are likely to carry less risks than M&A's. 

Even if an airline would choose focus as its basic strategy, there are pressures in the 
competitive environment suggesting the utilisation of alliances. Whether the airline 
bases its strategy on different customer groups (e.g. business travellers) or on certain 
geographic area (e.g. traffic between Europe and South America), it is nevertheless 
likely to benefit from some sort of partnership with suitable airlines. The simple 
rationale is that no matter what the niche or specific geographic market is, an airline is 
likely to benefit from a larger catchment area and better connections. 

As to the airlines choosing the lowest cost strategy, alliances may be of lesser 
importance, at least in the light of today's experiences from the nature of operations 
by low-cost carriers. In Europe the low-cost airlines, in practice charter carriers, 
typically cater for tourist traffic in and out of hoHday destinations, and in this type of 
traffic connecting flights provide only limited value added. Elsewhere in the world, 
primarily in the U.S.A., low-cost carriers serve the business traveller segment, too, but 
so far a major part of the business has been on domestic point-to-point markets where 
the value added provided by good connections is not necessarily essential. Here it is 
necessary to make a distinction between low-cost carriers such as Southwest Airlines 
and the feeder carriers, such as American Eagle. However, the fact that today alliances 
of low-cost carriers are rare should not be interpreted so that there is no potential in 
building an international alliance of low-cost carriers for the ever-more important 
leisure travel segment; in fact that might provide interesting opportunities in the ever 
more olobal tourist market of the future. 



o' 



As to the drivers of international alliances, it appears that, first, the changes in the 
industr)- have made it essential for most carriers to seek growth and to secure presence 
in a larger market; second, the many types of regulation in the industp.' make alliances 
the only feasible way to grow and seek presence in a larger market; and third, there is 
a pressure to utilise resources better, i.e. to reduce the operating costs. 



Getting around various forms of regulation in the airiine industry is a major motivator 
for alliances. For one thing, the fact that governments still are notable owners in many 
internationally operating airlines makes acquisitions of and mergers with airlines 
somewhat problematic in general - namely, national flag carriers are still considered 
in many countries part of national property and country image, and therefore foreign 
ownership is not seen favourably. Secondly, governments have often set specific 
limitations on the share of foreign ownership in the airlines of their nationality. 
Thirdly, antitrust legislation makes mergers and acquisitions problematic in many 
countries because these often would lead to a dominant if not monopoly position of 
the united firm at least in some markets; the background for this is that for historic 
regulatory reasons there are very many markets where a duopoly exists. Finally, the 
fact that bilateral agreements between countries still form much of the basis for 
international air transportation causes some problems. 

Securing market presence can be seen either as an offensive objective of alliances, 
typical for large airiines, or as an defensive objective, typical for medium-sized and 
small airlines. Larger airlines seek market power and consequent enhanced value for 
customer by pursuing larger network coverage, higher frequencies, more extensive 
loyalty programs and dominance of so-called hub aiiports through alliance 
arrangements. Medium-sized and small carriers appear to seek more market coverage 
rather than outright market power in order to respond to the challenge by expanding 
larger airlines; smaller carriers seem to consider participation in alliances essential in 
trying to avoid shrinking into mere regional operators - which, of course, might be the 
destiny of small carriers even with alliance arrangements. In the market presence 
objective of airline alliances it is necessary to distinguish the global level and the 
specific market level, which may require arrangements of conflicting interests. For 
example, for many reasons it is valuable for SAS to co-ordinate closely its operations 
on a global level with its Star Alliance partners, but in the specific markets of the 
Nordic countries SAS may need to deviate from the ideal Star Alliance strategy 
because it needs to respond decisively to the challenges by Finnair, a key rival in the 
home market of SAS. 

The third motivator for alliances is the need to utilise resources better. This can be 
pursued either through higher productivity or simply lower costs. Higher productivity 
is sought, for example, through sharing aircraft and air crew capacity, using partner's 
ground handling and airport passenger services at foreign stations instead of providing 
them by the airiine itself, and making better use of possible excess aircraft 
maintenance capacity by servicing partner airlines' aircraft. Capacity sharing 
arrangements can often be complemented by specialisation; for example, one partner 
can specialise in the maintenance of aircraft engines from a certain manufacturer and 
another partner in engines from another manufacturer. As to direct cost savings, for 
example joint sourcing of fuel, catering (food), aircraft, spare parts, or information 
and marketing services may produce significantly lower costs than sourcing alone by 
each partner. 

.As to the relative role of the three alliance motivators it appears that it is the pursuit of 
stronger market presence that clearly has been more apparent and dominant, the need 



for better resource utilisation being secondary and also more of a longer-term nature. 
Concerning the role of regulation and the need to circumvent it, the assessment is 
somewhat difficult. It seems very often to be a key factor in alliance building, part of 
the environment for nearly all airline cooperation. 

It appears that different drivers have a different nature, or perhaps justification, in 
different kind of airlines. In this respect airlines can be grouped roughly into large and 
small firms, and the relevant dimensions for the nature of drivers can be determined as 
tactical-vs.-strategic and defensive-vs. -offensive. Figure 2 illustrates the positioning 
of the drivers along these dimensions for large airlines, and Figure 3 for small airlines. 

(Insert Figure 2 about here) 

(Insert Figure 3 about here) 

It was found out in the study that the role of learning from partners in airline alliances 
is evidently quite insignificant. Very much differently from many manufacturing 
industries, where the ability to learn from a more experienced or otherwise better 
partner is often given as a reason for building alliances, in this study the factor hardly 
ever came up. Even if it is understandable that airlines are not very eager to publicly 
shout about their needs to learn from other airlines, thereby indicating their own 
possible deficiencies, it still appears that airlines generally speaking do not make 
sufficient use of the opportunity to learn better practices. 

Turning competitor airlines partners rather than rivals would appear to be a very valid 
motivator in today's airline business. This, however, hardly ever came up specifically 
along the study. Certainly it is true that firms are not keen to pinpoint their archrivals 
in an industry of such turmoil - where today's rival can be tomorrow's partner and 
vice versa - but nevertheless it was somewhat unexpected that something which is 
here called competitor taming was never. really suggested by the airlines. Earher 
research, mainly concerning manufacturing industries, has suggested that making 
friends out of foes would be a motivation for many alliances. What is called 
competitor taming here is very close to what Doz and Hamel (1998) have termed "co- 
option". Again, just like with regulation evasion, the airlines' rush for market 
presence and for resource utilisation perhaps just overshadows the competitor taming 
as a motivator for alliances, but presumably it is a hidden factor in many alliance 
cases. 

One outcome of the study is the evidence of the essential role that market presence 
plays in airlines' strategic planning for survival and prosperity - having global reach 
appears to be a must in most airlines' strategic plan. Also, it appears that resource 
utilisation is a factor often acknowledged but quite slowly actively pursued. One 
explanation for this slow action is, of course, the rigidity that airline management face 
due to both ver>' strong labour unions and regulation by authorities. However, overall 
it appears that the firms are rushing so hard to secure positions as to their market reach 
that they are paying perhaps too little attention to the longer term factor of learning 
from alliances. Maybe the histor\' of airlines as national icons, at least in Europe, have 
created corporate cultures that are not the best environments for absorbing new 



practices. It would seem that the crisis of the industry in the early 1990's has brought 
some more flexibility in many airlines, but a comparison to other industries would 
indicate that there is still quite a way to go, but a way with great potential. 

Managerial challenges 

It appears that alliance building is such a part of evolution in the airline industry that 
most airlines need to participate in - the opportunity cost of not participating might 
prove too high. Management in airlines face considerable challenges in making the 
alliances work: there is the pressure from authorities, demands by unions, perhaps 
mixed ownership by government and private parties, and the normal challenges of 
different cultures in different countries and firms, differing organisational 
arrangements in airlines, and strong personalities as airline executives. 

Managing the relationship with the governments and even local authorities would 
appear to be a major task for airiine management of the future. In addition to the 
national interests - country image, employment and balance of payment issues - the 
increasing role of ecological aspects (noise, pollution) will add to the importance of 
managing all sort of regulation. 

Outside regulation issues, it seems that the area where airiine management face 
hardest challenges and where there appears to be much potential for improvement is 
that of learning. Eariier research (Inkpen 1998, 225) has emphasised the role of trust 
between the partners as a contributor to successful leaming. It is the very notion of 
trust that makes airline alliances different from those in many other industries: so far 
alliances in the airline industry have been either very short lived or limited in scope, 
or both, and therefore the trust has not been developed between the partners. 

SUMMARY AND CONCLUSIONS 

Changes in the industry have made it essential for most airiines to seek growth and to 
secure presence in a larger market. There is again the pressure to reduce operational 
costs in airlines through better utilisation of resources. Finally, the many types of 
regulation in the industry make aUiances the only feasible way to grow and seek 
presence in a larger market. 

Airlines appear rather frustrated by the fact that the natural evolution towards 
transnational companies in this industry is effectively blocked by authorities. This 
frustration is echoed in the comment by Paul Moore, spokesman for Virgin Atlantic, 
concerning U.S. limitations for foreign airiines on acquiring or setting up a U.S. 
subsidiary: 

It's blatant protectionism. Alliances are an artificial solution to an artificial 
problem. There is no reason why the rules should not be different now. 
(Airline Business. October 1998, p. 76) 
As to governments' role in regulating the formation of truly transnational airiines, it 
appears to be a question of finding the right balance between enough freedom to allow 
efficiency in the global airiine industr>' to develop, but enough regulation to make 
sure that there is competition between the alliance groups at least in most markets. 



Market presence appears to play an essential role in airlines' strategic planning for 
survival and prosperity - having global reach appears to be a must in most airlines' 
strategic plan. Therefore the primary motivation for international alliances so far has 
been the need to secure an extensive catchment area or a large onward connection 
network. 

It seems that resource utilisation is a factor that is very often acknowledged in 
international airline alliance arrangements. However, airlines have in fact been rather 
slow in pursuing higher productivity or outright lower costs through concerted efforts 
with partners; there seems to be rigidity in airlines in operationalising the changes and 
therefore the resource utilisation has so far not been as significant a motivator as 
market presence. 

Overall it appears that the pursuit of stronger market presence has been more apparent 
and dominant, the need for better resource utilisation being clearly secondary and also 
more of a longer-term nature. The role of regulation and the need to circumvent it is 
rather difficult to assess. It seems very often to be a factor in alliance building, but 
very rarely is it expressed as the primary reason. However, the demands from the 
airlines to allow more freedom to rationalise and restructure the industry may get 
more outspoken in the future. 



REFERENCES 

Air Transport Intelligence (1998): Survey on airline alliances, reported in Airline 
Business, June 1998, pp.42-81. 

Airline Business (1990): Strategic illusions, pp. 24-30. 

Alamdari, F. and Morrell, P. (1997): Airline alliances: a catalyst for regulatory 
change in key markets? Journal of Air Transport Management, vol.3, no.l , pp. 1-2. 

Annual reports from 1988-97/98 from the following airlines: Air Canada, American 
Airlines, British Airways, Canadian Airhnes (PWA Corp.), Delta Air Lines, Firmair, 
KLM, Lufthansa, Qantas Airways, SAS, Swissair, Thai Airways International, United 
Airlines. 

Antoniou, A. (1998): The status of the core in the airline industry: the case of the 
European market. Managerial and Decision Economics, 19, pp. 43 54. 

Bartlett, A. and Ghoshal, S. (1995): Transnational Management - texts, cases, and 
readings in cross-border management. 2'"" edition. Irvvin, Chicago. 

Bissessur, A. (1996): The identification and analysis of the Critical Success Factors 
of strateeic airline alliances. PliD dissertation, Cranfield University. 



Borys, B. and Jemison, D. (1989): Hybrid arrangements as strategic alliances: 
theoretical issues in organizational combinations. Academy of Management Review 
14:2, pp. 234-249. 

Brouthers, K., Brouthers, L. and Wilkinson, T. (1995): Strategic alliances: choose 
your alliances. Long Range Planning, Vol. 28, no. 3, pp. 18-25. 

Button, K. (1997): International airline safety in a world of alliances. Conference 
Proceedings of the 1997 Air Transport Research Group of the WCTR Society, vol. 3, 
nr. 1. University of Nebraska at Omaha. 

Chan, S., Kensinger, J., Keown, A. and Martin, J. (1997): Do strategic alliances 
really create value? Journal of Financial Economics, vol. 46, pp. 1 99-22 1 . 

Contractor, F. and Lorange, P. (eds.) (1988): Co-operative strategies and 
international business. Lexington Books, Lexington, Mass. 

Doz, Y. (1996): The evolution of cooperation in strategic alliances: initial conditions 
or learning processes.? Strategic Management Journal, special issue 17, summer 1996 
pp. 55-84. 

Doz, Y. and Hamel, G. (1998): Alliance Advantage: the Art of Creating Value 
through Partnering. Harvard Business School Press, Boston. 

Dussauge, P. and Garrette, B. (1995): Determinants of success in international 
strategic alliances: evidence from the global aerospace industry. Journal of 
International Business, 26:3, pp. 505-530. 

Gallacher, J. (1998): A time for celebration. Airline Business. September 1998 pp 
29-58. 

Gant, J. (1995): The Science of Alliance. Euro Business, September 1995, pp. 70-73. 

Gleister, K. and Buckley, P. (1998): Measures of performance in UK international 
alliances. Organization Studies, vol. 19, nr. 1, pp. 89-1 18. 

Gulati, R. (1995): Does familiarity breed trust? The implications of repeated ties for 
contractual choice in alliances. Academy of Management Journal, 38, pp. 85-112. 

Hamel, Gary (1991): Competition for competence and interpartner learning within 
international strategic alliances. Strategic Management Journal, 12, pp. 83-103. 

Hamel, G., Doz, Y. and Prahalad, C. (1989): Collaborate with your competitors - 
and win. Han>ard Business Review, 67, pp. 133-139. 

Inkpen, Andre^v (1998): Learning, knowledge acquisition, and strategic alliances. 
European Management JournaL vol. 16, no. 2, April 1998, pp. 223-229. 



Killing, J. (1988): Understanding alliances: the role of task and organizational 
complexity. In Contractor, F. and Lorange, P. (eds.). Co-operative strategies and 
international business, pp. 55-67, Lexington Books, Lexington, Mass. 

Lei, D., Slocum. J. and Pitts, R. (1997): Building cooperative advantage: managing 
strategic alliances to promote organisational learning. Journal of World Business, vol. 
32,nr. 3, pp. 203-223. 

Lindquist, J. (1996): Marriages made in heaven? The Avmark Aviation Economist, 
Vol. 13, no. l,pp. 12-13. 

Lug, Y. (1996): Evaluating performance of strategic alliances in China. Long Range 
Planning, Vol. 29, no. 4, pp. 534-542. 

Ohmae, K. (1989): The global logic of strategic alliances. Harvard Business Review, 
March/April 1989. 

Ott, J. and Sparaco, P. (1997): Unique industry facets shape Europe's alliances. 
Aviation Week and Space Technology, Nov. 17, pp. 63-65. 

Park, J. and Zhang, A. (1997): Effects of intercontinental alliances: cases in the 
North Atlantic market. Conference Proceedings of the 1997 Air Transport Research 
Group of the WCTR Society, vol. 3, nr. 1. University of Nebraska at Omaha. 

Park, J. and Zhang, A. (1998): Strategic alliance and firm value: a case study of the 
British Airways / US Air alliance. Paper presented at the 1998 Air Transport Research 
Group of the WCTR Society Conference. 

Park, N. and Cho, D. (1997): The effect of strategic alliance on performance: a study 
of international airline performance. Conference Proceedings of the 1997 Air 
Transport Research Group of the WCTR Society, vol. 1, nr. 1. University of Nebraska 
at Omaha. 

Parkhe, A. (1991): Interfirm diversity, organizational learning and longevity in 
global strategic alliances. Journal of International Business Studies. 22: 5, pp. 579- 
601. 

Parkhe, A. (1993): Strategic alliance structuring: a game theoretic and transaction 
cost examination of interfirm cooperation. Academy of Management Journal, 36:4, 
pp. 794-829. 

Pekar, P. and Allio, R. (1994): Making alliances work - guidelines for success. Long 
Range Planning, vol. 27, no. 4, pp. 54-65. 

Phatak, A. (1997): International Management. South-Westem College Publishing, 
Cincinnati. 



Rhoades, D. and Lush, H. (1997): A typology of strategic alliances in the airline 
industry: propositions for stability and duration. Conference Proceedings of the 1997 
Air Transport Research Group of the WCTR Society, vol. 1, nr. 1. University of 
Nebraska at Omaha. 

Seristo, H. (1996): An executive view on the cost problem of European airlines. 
European Business Review, vol. 96, no. 4, pp. 14-17. 

Seristo, H. (1995): Airline Performance and Costs - an Analysis of Performance 
Measurement and Cost Reduction in Major Airlines. Doctorate thesis, Helsinki 
School of Economics and Business Administration Press, A- 107, 254 pages'. 

Seristo, H. (1993): Strategies of Airlines in the Deregulated European Competitive 
Environment. Helsinki School of Economics and Business Administration Press, B- 
136, 213 pages. 

Seristo, H. and Vepsalainen, A. (1997): Airline cost drivers: cost implications of 

fleet, routes, and personnel policies. yo»r«a/o/^z>rran5po/-/A/a«aFeme/i/ vol 3 no 
1, 1997, pp. 11-22. ' ■ ' ■ 

Stiles, J. and Turner, 1,(1994): An investigation into the strategic determinants of 
partner selection in international strategic alliances in the service sector. Working 
paper No. 94/1, Henley Management College. 

Youssef, W. (1992): Causes and effects of international airline equity alliances PhD 
dissertation, series UCB-ITS-DS-92-1, Institute of Transportation Studies, University 
of California, Berkeley. 





cr 


0) 


0) 


u 


> 


a 


•w 


a 


U 




(U 






< 


o 





V) 


k< 


c 


o 


o 






a 


•t.' 


S 


o 





cr 




(U 


>> 


> 


WD 


-*-• 


CJ 


rt 




c 


C/^ 


■^-i' 




CS 



C5 O 



^ 



•"■ 


C/5 


CA 


c« 


« 


U 


OJD 


c 


C 


'c« 




•^ 


U 


J2 






u 


IS 


s 


t5/> 




^ 


C/1 


n 






u 


*"*^ 


Oi 


(U 


•«^ 


> 


0) 




.:^ 


r: 


k. 


.iJ 



c 

e 

u 
o 

c 

03 



So 



> 

JS A A 



ail 

on ■- 



c 

« y 
00 c 

o 

c 



> 
o 
o 



(U <L) 



u 

3 



"T: : in C3 -3 



^ E 



'S A A 
Q ' ' 



e 

e 
o 



< 



o 



T3 



tn 



w 



S C o 



^ 3 OT 

»S C3 1 



en 

c 



a > 
■•s " 

— S " 



V 



u 



o 

S ^ '^ 



'o 

c 





<u 




n. 




E 




o 


DJ) 


CJ 


a 


:5 


s 


u 

c 


n 


A 


u 


1 



( ,__ 



B S3 ■'-' 

.5 S E 
= c a 



C3 

e 



e 









•a 



p. 

O 

(U 

O 



o 
O 



o 

o 



1 e^. 




I an 




1 ° 








! S 




« 




I -4-i 


, 1 


■ b. 


w , 


■ 


> j 


1 ^S 

1 '4-' 


^ ! 


1 U 


3 ' 


0. 


w ' 


1 E 


A ! 


1 




' U 




t 





u 








60 




03 


c 


C 


«. 


C3 




> 


t^. 


•a 





03 


^_ 


4-> 


n 


M 


> 





& 





■ — ' 


U-, 


t: 





c 




«; 


P 




l-> 


K 


3 


c 





t/l 


M 


c 





■♦^ 


JIl 


V 




cd 


c 


cr 





g. 




u 


-a 




u 


c 


00 


.c 


C3 


C 


^ 


-4— » 




c/: 




OJ 


1 - 


^ 


CO 










J 





« 


^ 





ch 


M 









■a 


-0 




<— 







Q 









1) 
tn 



c 
c o 



>% C3 






o 



G 

o 
>^ '-5 






CO 



o 
u 



5 i: 

C ° 
o cr 

< 



CJ 





C 





3 


f > 




L« 




ri 


n 




U 






o 



C3 

3 o 

w to 



5 

CJ 



c: ^ 



o 

3 S 

5 .2 

II 



DO 

5 



Q 



CO 

_o 

o 

c 

73 
o 



o 

CJ 

>-» 

(DJD 

■*-> 
d 
u 
<«^ 



u 

o 
u 

a 

C3 



en 
O 



a; 









■4— • 












C 












(l> 












E 








CO 




u 






o 


1) 
> 
O 

o 


4) 


u 

c 


U5 

Oh 

tl-c 


C 


a 


^ 


o 

c 

3 
1 

1 


<u 




f 


1 


1 

c 

1 


> 

A 

1 


I 

1 


o 

3 
1 



> 

• l-H 

c 



V3 

;^ 

3m 

cs 
u 

a; 
W) 



i2 «« .s 

O Si .JT M I 



ii .w I 



> P « -5 
c "* ^ 'ci 

2 fc §) <; I 





^ 










B 


> 




O 






R 

ce 


o 

3 


■iS- 


S 




O 

o 


u 

a 






3 




o 


"O 








A 


A 


Pi 







V3 



C3 

a 
e 



Of 



c 

R 
W 



bJi 



e 

R 



o 



c 

B 

o 

U 



o 

'5b 
o 



'55 

c 

tt-l 

Q 



o 

1/5 

^> 



f 

3 






00 



o 



I 



o 



.2 
'o 
u 

00 





CO 1 




(U 1 








O 1 


T 


■n-i 




<D 1 








<U I 




O.I 




£ 1 




o ' 


1 (5J) 


u " 


' C 


& ! 


1 s 


OJ , 


h. 


c I 


1 4, 


^ ' 


1 ^ 





u ex 






u 



>■ c " « 

c "^ I 'g 

« S3 &<; 

Ph I I I 



u 

> 

O 

o 



o 






5 c 



c 
<u 

E 
c « 

u ^ fa 

§ ^^ 

HJ > ^ 
^^ A — 



OX) 

B 

is 

•- w 

U >-< 



<D 
> 

c 





>% 






^-» 




e 


> 




o 












■«^ 


o 




R 


3 




S 


O 


w 
O 
o 


w 


(l> 






on 


o 

> 


s 
o 


j=. 


_o 




A 


A 


c^ 







c 

M— I 
CD 

Q 






I 

c« 

u 

O 

.2 

a 

13 

a 
o 

■*^ 

u 



o 



f 



o 

"5b 

00 



C3 

o 



Air Transport Policy in Japan: 
Policy Change and Market Competition 



Hirotaka YAMAUCHI, 

Professor, Faculty of Commerce, Hitotsubashi University. 
Tokyo, Japan 

February 1999 



1 Introduction 

Japanese air transport market developed in a strictly regulated environment. The 
Civil Aeronautics Law, which governs the industry, requires that firms should obtain 
government licenses to get into the market. Airlines also need government approval for 
their fares, and even for their annual business plans. 

But a policy stream toward liberalization of air transport since the 1980s has brought 
to fruition of substantial deregulation in this market. Now the Ministry of Transport 
(MoT) is going to submit a bill to the Diet, which revises operating licensing system, fare 
approval system and other regulatory provisions. Moreover, since the MoT relaxed it's 
operating standard of administrative process, we can say that now the airline industry is 
under competition. As widely reported, newly established carriers entered into markets 
and their impacts on market competition were strong although shares of new companies are 
quite small. 

As for international air transport, liberalization has been also in advance. The 
Memorandum of Understanding concluded on March 14 1998 with the U.S. government 
was an outcome of negotiation to equalize the right and interests between two countries, 
but its essential factors are thought to be giving airlines of both countries with freedom to 
conduct in the market place. And strategic alliances among world airlines will make it 
more severe to compete in international markets. 

Facing with the new stage of competition among air carriers, we have to move on a 
new air transport policy, which should pursuit efficiency and fairness in this market. 
Needless to say, efficient air transport system is the infrastructure of sound economic 
development and globalization of economy. Tji >l-iit r-nr-^ jt [^ tlr:mglit t':' I": " vr - 

■£aciiiuij.? ffficiirnt air transport market in toopcration viith oncli othci'. 

2 DEREGULATION OF THE DOMESTIC MARKET 

2.1 Abolition of Supply-Demand Balance Clause 

In Japan, we are now facing a powerful policy trend reconsidering the role of the 
government in economic policy, and there are emerging consensus that deregulation is the 
only way to revitalize the economy as a whole, to recover international competitiveness 

1 



and to beneTit consumers. The transport field is not an exception. 

On December 5 1996, the MoT announced that it would abolish supply-demand 
balance clauses in every transport business law including the Civil Aeronautics Law in a 
couple of years. A supply-demand balance clause provides that a new entry or increase of 
supply by existing carriers could be approved if and only if the MoT make a judgement that 
the balance of supply and demand in the market would be disturbed. This is a typical 
quantitative control of supply and the clause might effectively block new entry. 

The^eattee has also given the MoT with wide range of administrative discretion, 
because, according to the clause, it is not the company managers but the government 
officials that judge whether there is excess demand or not. The abolition of the clause 
means that that there could emerge much room for effective competition carriers than in 
present situation, since managers become able to make decision on their own judgement 
and to take actions timely. 

2.2 Industrial Policy in Air Transport: an Old Regime 

As noted above, Japanese airlines were fostered in strictly regulated environment. 
The governmental intervention in this industry was conducted not only through statutory 
actions but also through purely administrative process such as a cabinet meeting resolution 
and a notice from the Minister of Transport. Especially, the Cabinet Meeting Resolution 
in 1970 and the Notice from the Minister of Transport in 1972 played a role to fix the 
market structure in the air transport and some times called "Aviation Constitution". The 
Civil Aviation Law and administrative guidance did not allow airlines from competing 
rigorously. 

This "old regime" was intended to secure and nurture the capacity of all members of 
the airline companies by establishing segmented business fields for each firm. The 
segmentation of market was also a common feature of Japanese industrial policy in 1950s, 
60s and first of 70s. In air transport case, routes licensing regulation could make the 
segmentation concrete and trunk routes markets offered a base for operational stability and 
became source for cross-subsidization. 

The old regime survived until mid-80s, with all three firms growing steadily within an 
arranged business base. The air transport market as a whole grew rapidly with a help of 
Japanese high economic expansion, and the route network was widened. The role of 



governmental intervention in the form of protection of infant industry can be said to have 
functioned adequately up to this stage. But the most serious problem of such a cartel- 
oriented government policy was that the high cost nature of airlines was bought about by 
protection from competition and that it remained even after the situation was changed. 

2.3 Policy Change in the Last Decade 

The old air transport regime collapsed in mid-80s. The trigger was the conclusion of 
the Japan-U.S. Aviation Treaty Interim Agreement of 1985 and the signing of its 
Memorandum of Understanding. The strategy of the Japanese government in the 1970s 
was to limit international schedule carrier to Japan Airlines (JAL), but, the Interim 
Agreement admitted the new entry of Nippon Cargo Airways (NCA), moreover it allowed 
other new carriers of both Japan and the U.S. to start scheduled passenger services. 
Naturally, to make this possible, it was necessary for the government to end JAL's 
monopoly over scheduled international service. Around this time, calls for the 
liberalization of the Japanese domestic air industry was also strengthened, and the Council 
for Transport Policy (an official advisory committee to the Minister) announced its opinion 
that the Old Regime formed in the first half of 1970s should be abolished, and that more 
pro-competitive air transport policy should be pursued. The content of its detailed advice 
were as follows: 

(1) International routes would be served by multiple carriers; 

(2) Competition on domestic routes would be promoted by new entry into particular 
city pair markets; and. 

(3) Japan Airlines would be completely privatized. 

The goveniment insisted that domestic aviation has moved onto a more competitive 
situation, because of the new aviation policy adopted in 1986. However, the system has 
met critics that the government's regulation of fare approval and entry licensing has 
basically remained unchanged, so even though several carriers compete over the same 
routes, these routes are subject to an entirely uniform fare structure. 

In response to such critics, the government adopted a policy that makes it easier to 
offer discounted fares in 1995 and a zone-fare system in 1996. This zone fare system 
adopted is similar to that adopted by European Communities (now European Union) before 
the third package of common air transpon policy was implemented in 1993. The system 



involves establishing a fixed price range and allowing carriers to set their air fares within 
that range at their own discretion. Needless to say, this allows carriers to respond to a 
particular demand period with a flexible fare structure. Carriers can introduce and set all 
types of discount fares, including advanced purchase fares, to meet the demand of different 
periods. 

The upper limit of the permitted fare zone is initially calculated based on the airlines' 
cost level. The lower end of the rage is set at 25% less than the upper limits for normal 
fare. The carrier can set discount fares at a maximum of 50% below the lower limit. 
Logically, the deepest discount fare could be set at 62.5% off compared with the upper 
limit fare. 

3 Competition in the Domestic Market 

3.1 Demand Structure 

The five-year growth rates in the number of air passengers are 9.7% (1975-80), 1.6% 
(1980-85), 8.3% (1985-90), and 3.7% (1990-95). Demand is periodically hampered by 
the capacity of Haneda Airport, which expanded in July 1988 by the New A runway and 
again, in 1998, the New C. Given Japan's geographical size, the air transport market in 
Japan is not small. The revenue passenger kilometers (RPK) in the domestic market total 
about 65 billions, one-tenth the U.S. figure, and 78 million passengers fly domestic routes, 
which is one-sixth the size of the U.S. market (1995 data). 

Using time-series data from 1974 to 1995, 1 estimate the aggregate demand function 
as follows: 

Ln{RPK) =10.157- 0.74 1 Ln{RFARE) + 1 .292Ln(RGDP), adjusted /?^=0.982, 
(5.430) (-3.665) (12.782) 

where RPK = revenue passenger kilometers, RFARE = real airfare (domestic yields per 
RPK, deflated by the CPI), RGDP = real GDP. '. 

Simple aggregate demand function analysis indicates that the long-term price 
elasticity of domestic air travel is about -0.74 and the long-term income elasticity is about 
+ 1.29. Compared with Ohta (1981), that suggested comparable figures of -083 and +1.66, 
my estimate shows an income elasticity decrease owing to the nevs'er data set. 



I estimatsd several other functional forms, including a dummy variable for fatal accidents, but the simplest 



The most important feature of this demand is the concentration on Tokyo routes. As 
shown in Figure 1, Haneda Airport handles about 55% of total air passenger in Japan, 
although the number of routes originating or terminating there only account for 17.9% of 
all routes. This is because many dense markets are involved, and revealed in Figure 2. 
Armual traffic on the Tokyo - Sapporo route is 7.6 million passengers, which is the largest 
in the world, and for Tokyo - Fukuoka route the figure is 6.2 million, which also ranks high 
in the world. The only non-Tokyo route ranked in the top ten domestically is Osaka - 
Sapporo.2 These demand features highlight the importance of operating rights at Haneda 
Airport, especially in view of the high-cost nature of Japanese air carriers. 

3.2 Carriers 

Eight scheduled airline companies operate in Japan. JAL, All Nippon Airways 
(ANA), and Japan Air System (JAS), these are the three earliest. Japan Asia Airlines 
(JAA) and Nippon Cargo Airways (NCA) offer only international service. Japan Trans- 
Ocean Airlines (JTA) and Japan Air Commuter (JAC) are solely domestic carriers, and Air 
Nippon (ANK) is mainly domestic but recently opened an international route between 
Fukuoka and Taipei. Note that JAA and JTA are subsidiary of JAL, NCA and ANK are 
of ANA, and JACis of JAS. 

In 1998, newly established two carriers entered into the domestic market, which are 
Skymark Airlines and Hokkaido International Airlines (Air DO). Skymark operates only 
in Tokyo - Fukuoka route and Air DO in Tokyo - Sapporo. These carriers very small and 
flights are very few, but since their fare strategy is very aggressive (very cheap), they 
succeeded in getting popularity. 

The largest is JAL, and in 1995 it carried about 72.4 billion RPK in domestic and 
international markets. This is one-half or one-third the figure for U.S. mega-carriers. 
The second and third largest are ANA (about 43.8 billion RPK) and JAS (.13.7 billion 
RPK). When the Japanese economy was booming, the airlines made big profits, but along 
with recession have come big deficits. The air transport market is becoming stable, but 
the airlines are restructuring. 



one fit the best. 

- The top three U.S. markets are berween New York and Los Angeles, Chicago and Washington, D.C., each 
with annu.il passengers of between 2.5 and 2.7 million. 



As noted earlier, one regulatory objective has been cross-subsidization between trunk 
routes and local routes. The extent is unknown because profit and loss accounting by 
routes is not reported to the public, but it is said that two-thirds of JAS's routes post losses 
and that is why it objects free entry and exit policy. It is claimed that unprofitable routes 
would be abandoned, and passengers without substitute transport modes would suffer. 
From an economist's point "of view, however, the solution for that would be to maintain 
service by general government subsidy. The United States has the Essential Air Service 
Program, and the EU's third package of Common Air Transport Policy has similar program. 
The Japanese government is now seeking for a new direct subsidy system to be 
implemented at the next stage of air transport liberalization.^ 

3.3 Market Structure 

Under the old regime, ANA had a major share in domestic market, but Figure 3 shows 
a decline form 57.4% to 47.2%. In a sense, this resulted from market liberalization, but it 
should be noted that none of ANA's competitors increased it share dramatically. Rather, 
each gained a few percentage points, while ANA's subsidiary, ANK, increased its share by 
2.5 percentage points. ANA transferred unprofitable routes to ANK to make its financial 
position healthier. The government policy adopted in the mid-1980s has not led to radical 
change in market structure. 

ANA has not lost share dramatically because of a strong sales network and brand 
loyalty in domestic the market, which were nurtured under the old regime. Furthermore, 
until very recently, fare competition was not allowed, and new entrants had no effective 
means to fight incumbents. In a sense, this also is a legacy of the old regulatory 
environment. 

Another reason shares have not changed is airport limitations. As stated earlier, 
Haneda is the biggest profit center for carriers but does not have enough capacity. 
Landing slots have not increased much, although the expansion project is underway. In 
such a situation, incumbents have a competitive advantage over new entrants but not other 
incumbents. 

Structure has changed in terms of city-pair markets with multiple carriers. Figure 4 



^ In April 199S, the Council for Transport Policy submitted a report on further liberalization of the air 
irnnsport nunrket, in which it was proposed that a new subsidv program should be established. 

6 



shows changes in the percentage of passengers by market type: single, double, and triple 
trucking routes. After the policy change, passenger traffic on multiple routes increased 
steadily, reaching to about 72% in 1994. This means that the majority of passengers 
could choose their carrier. But, as stated above, carriers did not have flexibility in setting 
fares and even passengers with choice of two or more airlines did not enjoy the benefits of 
competition, as there were no difference in service or prices. 

3.4 Airfare Trends 

The trend in average domestic airfares since the mid-1970s is shown in Figure 5. 
The average is calculated by dividing total revenue by total passenger kilometers for all 
carriers. Until recently, domestic airfares were tightly regulated, and the average 
remained relatively stable at least in nominal terms during the 1980s, after a hike in 1980 
due to the second oil crisis in the previous year. Stability in nominal terms generally 
means a decline in real terms. 

We can identify the dowiiward trend in airfares since 1990 in nominal as well as real 
terms. In this period, fares were still regulated, but carriers could offer travel agents 
discount fares for inclusive tour programs, which might be used illegally for seat sales. 
So we cannot deny the possibility that the downward fare trend in 1990 reflects entry 
relaxation in the mid-1980s with a time lag. But it should be noted that the Japanese 
economy was in depression, and the fare decrease could be due to the weak economy. In 
any case, air passengers did not realize benefits from competition, and this led to demands 
to relax fare regulation. 

The zone fare system was introduced in June 1, 1996, and in spring 1997, MOT 
reported on a comparison of the average fare with the previous year. As seen in Figure 6, 
the average fare declined by 2.3% in nominal terms. Since general consumer prices 
remained fairly stable during this period, this can be regarded as a real price decrease. 
The reduction is not trivial, considering that the annual rate of decline in U.S. domestic 
airfare since deregulation is 2.8% in real terms.'' It is not clear that this price decline was 
mainly due to the new zone system. The average domestic fare had started to decrease 
since 1990, and the drop between 1994 and 1995 was 3.5% in both nominal and real terms. 



■• According to Air Transport Association data, average U.S. airfare in 1977 was 13.4 cents per passenger 
mile, which declined to 8.07 cents in 1995 (calculated in constant dollars based on 19S2). 



Judging from aggregate data, domestic airfares in Japan have declined at a nontrivial 
rate, but consumers do not perceive much change. The main reason for their feeling 
seems to be that the absolute level of airfare in Japan is higher compared with that of other 
countries, especially the United State. 

4 REVOLVE OF INTERNATIONAL AIR TRASNPORT 

The operation of international air transport is based on bilateral agreements, which 
reflect reciprocal rights and interests of each country. Owing to protection of rights and 
interests, the negotiated traffic level is likely to be lower. The country with less 
competitive and less efficient carriers may well try to protect its airlines and to limit the 
within which its carriers can compete safely. 

A cartel initiated by the International Air Transport Association (lATA), stabilized 
international airfares and avoided substantial competition. Although lATA still exists and 
the Traffic Conferences of lATA are held regularly to set fares route by route, its ability to 
contain competition has been reduced. Its main role has shifted to cooperative functions, 
such as a debt and credit-clearing house for airlines. The degree of competition in 
international markets depends on the bilateral agreement, especially the capacity control 
clause. 

The Japanese government was persistently taken a rather traditional stance on 
international aviation negotiations, a tuming point in 1986 when the Council on Transport 
Policy submitted a report that suggested a new direction. The background to this report 
was the provisional agreement .with the United States made the previous year, which 
allowed more Japanese and U.S. carriers entering the market between two countries. By 
this agreement, ANA and JAS became international carriers, and United, American and 
Delta obtained access to Japan. 

Although this provisional agreement was not a liberal agreement giving carriers 
freedom in terms of capacity and price setting, it triggered changes in Japanese air transport 
policy. It was the starting point for relaxing entr>' conditions and expanding capacity 
expansion in international air transport markets. 

The reason the Japan-U.S. provisional agreement is not more liberal is that the 
Japanese government believes that there is an inequality of rights and interests in the 
Japan-U.S. bilateral agreement, and that this inequality hampered fair competition in air 

8 



transport market between two countries. The Japanese government insists that the follow 
inequality pertain. First, in the original agreement, the United State has unlimited fifth 
freedom rights beyond Japan, while Japan has only one point of that right beyond the 
United State. Second, the United State has more full right carriers than Japan. (Full 
right carriers can increase or decrease capacity without advanced notice.) Third, there is 
an imbalance in the capacity provisions in the north Pacific markets. Forth, as a result 
U.S. carriers can attain a greater share than Japanese careers in that market. 

Not all researchers agree with these assertions. It is pointed out that one cause of 
imbalances in capacity and market share is the failure of Japanese carriers to expand their 
capacity. It is true that there is an inequality in the beyond rights between two countries, 
but it is worthwhile noting that these rights are not so attractive to Japanese carriers 
because of Japan's geographical location. 

Generally speaking, the complaints from foreign countries regarding Japanese 
international aviation policy focus on the difficulty in entering the Japanese market and in 
increasing their capacity. These complaints are partly caused by Japan's policy, largely 
stemmed from airport congestion problems.' 

In March of 1998, Japan and US agreed a new memorandum of understanding. In 
the negotiation process, while US government strongly insisted that Japan should accepted 
the liberal agreement, since Japanese government refused it persistently, the MoU was not 
said to be liberal agreement. The official reason why Japan opposed liberal agreement 
was that there remained the inequality of rights and interests in the bilateral mentioned 
above. 

However, the essence of the MoU was to introduce greater competitive environment 
into the North Pacific market. The new agreement allows for full right can-iers to chose 
any city pair market between two countries if there is no landing slot problem, to exercise 
beyond right more freely than present^ and to take use of code sharing even between same 
country's can-iers. Moreover, the agreement equalize the number of full right carrier for 
two countries, which could meets Japan's complaint about inequality in the original 



- As for the detail discussion on US-Japan bilateral agreemeni. see Yamauchi and Ito (1996). 
*" While there remained preconditions on using beyond right for both countries, these conditions are not 
restrictive. 

9 



bilateral agreement while for the non-full right carrier, flight increase is allowed. The 
new agreement was concluded with substantial compromise of two countries, but it is sure 
that competition among carriers will increase and increased competition would benefit 
consumers as well as air carriers themselves. 

5 CONCLUSION 

In this paper, we have examined chang in Japan's air transport policy. Air transport 
industry is quickly metamorphosing, so the policy should keep pace with its object. It 
should be noticed that competition makes airline more efficient and more competitive, and 
that sound air transport system brings huge benefits to consumers and global economy 
consequently. So the ultimate purpose of the policy is nothing but enhancing the 
competition in this field. 

But it is true that there are several problems to be solved in promoting competition in 
the air transport market. 

First, in order to make the competition fair and workable, we have to make sure of the 
equal footing for the market competition. For example, since the congestion in airport 
would be main obstacle to new entry and strategic decision making, we have to invent 
transparent and efficient procedure to allocating landing slots. In the international context, 
the government aids to its flag carrier are the most controversial problem. In the EU case, 
they put the judgement on legitimacy of government subsidy into the hands of EU officials, 
but the process for judgement was not thought to be clear and persuasive. 

Second, the global alliances among carriers put us difficult problems. As noted in 
the text, alliances are likely to make competition more active and increase passengers' 
benefits. But there is a possibility that it fosters worldwide oligopoly in international 
aviation, and if so, we will need someone or some authorities to keep on watching the 
behavior of players. It does not seem to be easy for us to agree with each other on the best 
regulator in this matter. 

-Fir«rt-di5€.ad^-ft^-IU— ccaJjiri.v fh^ '^ir tfir^pnrt H;"n:-'n i H in fh'\t iir - m'^y ..,<11 ai:hi';^''r tl ir 
-U^o^c-scgiaining problem j 



10 



References 

Yamauchi, Hirotaka [1997], "Air Transport Policy in Japan: Competition under 
Regulation," in Christopher Findlay, Chia Lin Sien and Karmjit Singh eds. Asia 
Pacific Air Transport: Challenges and Policy Reforms. Singapore; Institute of 
Southeast Asian Studies. 

Yamauchi, Hirotaka, and Takatoshi Ito [1996], "Air Transport Policy in Japan," in Gary C. 
Hufbauer and Christopher Findley eds. Flying High: Liberalizing Civil Aviation 
in the Asia Pacific. Washington DC; Institute for International Economics. 



11 



Oiher than Haneda and 
Itanii 
24% 




Between Haneda and 
tiami 
4% 



Figure 1. Passenger Shares of Haneda and Itami Airport 



Number of routes 







■ Other routes 
□ Haneda routes 






".- w^'T^*" 










■ ■'-■-•'..'■••"-. •'-■■■.■■■ ■ ,■"■ ■'J/,i^^i^./■^'■■ 











0-100.000 100,000-200.000 200.000-500.000 500.000-1.000,000 1,000,000+ 

Number of annual passengers 



Figure 2. Route Structure of Air Transport in Japan 



I9S4 



1994 




50% 60% 



10% 20% 30% 

[□ANA hIaL dJaS QJTA hank. QJAC 



«0% 90% 100% 





ANA 


JAL 


JAS 


JTA 


ANK 


JAC 


1994 


47.2% 


26.7% 


19.9% 


2.5% 


3.2% 


0.5% 


1984 


57.4% 


23.3% 


17.2% 


1 .4% 


0.7% 


0.0% 



Figure 3. Changes of Share in the Domestic Market 




87 88 89 90 91 92 93 

DTripIe Truck Roul<:;> ■ Double Truck Routes D Single Truck Routes 

Figure 4. Passenger Share by Market Type 



94 



30 



:o 



15 



10 




7-1 75 76 77 7S 79 80 



!- S3 84 85 86 87 SS 89 90 i\ 92 9J 94 95 
- nominal term —»- real term 



Figure 5. Trends in Average Airfares, 1974-1995 



Note;: ualcul.ited ;is lui.il p;issi:ni;or ri^eniii; divided b> total revenue passenger kilotneterv 



Yen 



Figure 6 Domestic Average Airfare (in nominal term) 



20 



e 
i. 
o 



^^:!^A^i§Mi§^-^S^-^'^-^^'^ 















-*- Average Fare 
^~' Shifting Average 



^ -~ ^'. ''-'-"•■-"• ' ■■•'J>"'I—V>^'- ,""'•' - • v.'*!'. 



' ' ' ■ ' I 




^^■■vvvv'i^-^^vvvvvvvvvvvvvvvvvy 



Aviation Infrastructure Performance and Airline Cost: 
A Statistical Cost Estimation Approach 



Mark M. Hansen 

David Gillen 

Reza Djafarian-Tehrani 

Institute of Transportation Studies 

National Center of Excellence in Aviation Operations Research (NEXTOR) 

University of California at Berkeley 

May, 1999 



Corresponding Author: Mark Hansen 

107B McLaughlin Hall 

Berkeley CA 94805 

Phone: 510-642-2880 

Fax:510-642-1246 

e-mail: hansen@,ce.berkelev.edu 



This research was supported by the U.S. Federal Aviation Administration and the Intel 
Corporation. 



Aviation Infrastructure Performance and Airline Cost: A Statistical Cost Estimation 

Approach 
-Abstract- 

The relationship between the performance of the U.S. National Airspace System (NAS) 
and airline costs is examined by estimating airline cost functions which include NAS 
performance metrics as arguments, using quarterly data for 10 U.S. domestic airlines. 
Performance metrics that vary by airline and quarter are developed by applying factor 
analysis to seven '^derlying variables, including average delay, delay variance, and the 
proportion of flightsJwhich are cancelled. This analysis reveals that variation in the seven 
variables can be adequately summarized by three or fewer factors, which we term NAS 
performance factors. If three factors are used, they correspond to delay", *('ariability", 
and tlisruption". The first of these captures average flight departure and arrival delay. 
The second reflects the variance in delay, while the third is based on the incidence of 
situations in which operations become sufficiently irregular to require flight 
cancellations. In the two-factor representation, '(variability" and tlisruption" factors are 
essentially merged into an ^regularity" factor, while the one factor model blends 
ftregularity" with delay". When the NAS performance factors are used as arguments in 
an airline cost function, the tlisruption" factor is found to be positive and significant in 
the three-factor model, as is the ftregularity" factor in the two-factor model. No 
significant effect is found in the cost ftinction with one performance factor. Using the 
estimated two- and three-factor models, we estimate the cost savings that would result if 
the NAS performance levels in each observation were improved to the highest level 
found in our data set, and find annual savings to be in the $1.5-2 billion range. The 
estimates are fairly consistent with previous estimates of the cost of delay based on 
applying delay tost factors "to the number of minutes of aggregate delay. On the other 
hand, our findings suggest that the main linkages between NAS performance and airline 
cost involve irregularity and disruption rather than the quantity of delay minutes. 



1. Introduction 

The need to understand and quantify the benefits of public and private investments in the 
National Airspace System (NAS) has never been greater. On the public side, Executive 
Order 12893, published in 1994, requires the Federal Aviation Administration (FAA) 
along with other federal agencies to conduct systematic analysis of benefits and costs of 
all infrastructure investments involving annual expenditures in excess of $50 million. 
The analysis is to t^uantify and monetize benefits and costs to the maximum extent 
possible (FAA, 1998). "Moreover, the FAA Acquisition Management System, published 
in 1997, mandates an investment analysis" prior to the initiation of a new acquisition 
program, including, among other things, the identification of alternatives and assessments 
of their benefits and costs (FAA, 1998). Such analyses are required for a host of Air 
Traffic Management (ATM) and Communications-Navigation-Surveillance (CNS) 
programs through which FAA intends to modemize the NAS over the next two decades 
(FAA, 1999). 

Private investments, particularly those by airlines in advanced avionics for new aircraft, 
are also getting closer economic scrutiny. According to Allen et al (1998), the industry 
is getting to the point where the achievement of business case maturity may be more 
important than technical maturity." Business case maturity includes the ability to 
explicitly identify benefit mechanisms triggered by CNS/ATM investments, credible 
estimates of the dollar values flowing from these mechanisms, and explicit analysis of 
investment risk (Allen et al, 1998). The CNS/ATM Focused Team (C/AFT), whose 
membership includes airframe manufacturers, airlines, and the FAA, has been working 
since 1997 to develop and apply a methodology for developing such business cases. 

While the need for benefit quantification is growing, industry stakeholders are also 
recognizing that the performance of the NAS is multi-dimensional, and therefore not 
adequately captured by traditional, delay-based, metrics. For example, the C/AFT has 
identified six categories of performance, including, in addition to delay, predictability, 
flexibility, efficiency, access, and cost of service (Alcabin, 1999). These concepts are 
considered to tfefine the elements of value to the scheduled airline business "as well as 
the common criteria for developing economic models needed to predict 
benefits- •• (Alcabin, 1999)." 

Taken together, these trends suggest that NAS investment analyses should consider, and 
attempt to monetize, the impacts of a proposed investment on multiple dimensions of 
NAS performance. Unfortunately, the state of practice falls far short of this ideal. Almost 
without exception, investments analyses and business cases consider only delay and 
direct cost savings when evaluating the benefit of a NAS improvement. For example, a 
recent business case for advanced data link (ATS Data Link Focus Group, 1999), 
considered to be a path-breaking effort within the industry, identifies four benefit 
categories. Two involve communication cost savings, one is increased availability of 
communication between aircraft and airline operations centers (this was guessed to be 
worth anywhere between S16 and S48 per flight), and the last categor\' is delay cost 
savings (valued at S25 per minute based on aircraft direct operating cost). 



Thus, even as industry stakeholders recognize that NAS performance has many aspects, 
only delay is routinely monetized. Even here, however, there is ample room for 
skepticism about the procedures. Virtually all delay cost calculations involve nothing 
more than the application of a cost factor based on reported values for the average direct 
aircraft operating cost per block hour to quantities of delay measured in time units. For 
air transport aircraft, the cost factor is in the range of $20-$25 per minute. A few studies 
refine this figure by differentiating between delay taken at the gate, on the ground, and in 
the air (Odoni, 1995; Geissinger, 1989). Others extend the calculations by disaggregating 
expense by functional category, such a ftiel, flight personnel, maintenance, and capital, 
and estimating how delay, portrayed as changes in the quantity of block hours, affects 
each one (Kostiuk et al, 1998). 

The approaches to delay cost estimation share some strong assumptions that are rarely 
scrutinized or even acknowledged as such. These include that the cost of delay is an 
additive function of the cost of individual delay events, and that the cost of each event is 
a linear fiinction of the duration of the delay (and perhaps the phase of flight in which it 
occurs). Such assumptions ignore the possibility that delay cost is non-linearly related to 
duration, is subject to combinatorial effects, and includes sizable indirect components. 

It is probable that the cost of a delay varies nonlinearly with the duration of the delay. For 
example, one 40-minute delay is more costly than 40 one-minute delays. The 40-minute 
delay is far more likely to disrupt ground operations, gate assigrmients, crew schedules, 
and passenger itineraries. Conversely, airlines sometimes add delays to flights to, for 
example, avoid having a flight arrive at a hub in the middle of a departure bank. If this is 
rational behavior, then the relationship between cost and delay must not only be non- 
linear, but also non-mono tonic. 

Delay costs are also subject to combinatorial effects. The severity of the impacts noted 
above is likely to depend not only on the duration of delay to a specific flight but on the 
interaction of delays for many flights. This is particularly evident in a hub-and-spoke 
network in which flights are scheduled in connecting banks. If all the flights in an 
inbound bank are delayed by the same amount, then the effect may be far less severe than 
if half the flights are delayed by a larger (or even the same) amount. 

Finally, the prevalence of delays may generate sizable indirect costs through airline 
adaptation behaviors. Carriers may take a variety of measures to make their operations 
more robust to delay. These include building more padding into scheduled block times, 
providing flights with additional fuel, and having extra aircraft, flight crew, and ground 
personnel available. While these measures decrease the cost of delays when they occur, 
they also increase costs of day-to-day operation. In this way the cost of delay may 
permeate throughout the entire cost structure of the airline in ways that are not tied to 
individual delays events. 

Our limited ability to take these aspects of delay cost into account, combined with the 
nearly complete absence of information on how to place an economic value on other 



dimensions of NAS performance, represent critical gaps in knowledge at a time when 
massive investments in the system are being contemplated. One might attempt to fill 
these gaps in a variety of ways. Simulation is one possibility. To address the questions 
under consideration here, a simulation would have to be highly detailed. It would need to 
capture how airlines respond on a real-time basis to operational irregularities and the cost 
implications of that response. The problem gets especially complicated when major 
adjustments such as rerouting of aircraft and reassigning crews are considered. While 
such a simulation may eventually be possible, it is beyond our present capabilities. 

A second possibility is to systematically query airline personnel. For example, one might 
present dispatchers with different scenarios concerning the operation of their assigned 
flights throughout the day or month, and ask them to choose which scenarios are more 
desirable. If the scenarios were carefully chosen, this procedure would reveal the 
preferences of the participants, and thereby allow the estimation of utility functions 
whose arguments would be various dimensions of NAS performance. Such a study might 
yield very useful results, but is also subject to a number of objections. First, it is not clear 
how such a methodology could allow monetary valuation of NAS performance, since this 
would require participants to choose between scenarios that involve money as well as 
flight operations. Second, it is not obvious that dispatchers, or any other airline personnel, 
have a sufficiently global view of the airline 's interest to make the correct choices! 
Finally, the results of such a study might be biased by principal/agent effects, with 
respondents making choices that are best for them rather than for the airline as a whole. 

This paper focuses on a third approach, which is to estimate airline cost functions 
including NAS performance measures as arguments. Using published, quarterly, airline- 
level data, we estimate relationships between airline operating expense, outputs, factor 
prices, and other variables. Included among the latter are a set of variables, which we 
term NAS performance factors, " that quantify the airline 's operational experience in the 
NAS during the quarter. By observing how these variables influence airline expense, we 
establish a direct empirical basis for translating various dimensions of NAS performance 
into monetary terms. Any quantifiable aspect of NAS performance can, in principle, be 
accommodated in this framework. Moreover, because relationships are derived from 
observed co-variation between performance variables and cost, the results entail a 
minimum of assumptions about the mechanisms involved. 

This paper presents a first step in using cost estimation to assess the economic value of 
NAS performance. It employs a relatively small data set and, accordingly, a limited set of 
NAS performance variables and a simple form for the cost function. Nonetheless, it 
yields plausible results, including industry-wide estimates of the costs from Sub- 
optimal " NAS performance that can be compared with results of more conventional 
studies based on delay cost factors. This suggests that statistical cost estimation is a 
promising avenue for assessing the economic benefits of NAS improvements. 

We proceed as follows. In the next section, we present our analytical framework In 
Section 3. we describe and present results from our procedure for developing airline level 
N.AS performance variables. Section 4 turns to the specification and estimation of our 



cost model, which we then use, in Section 5, to estimate aidine cost savings from 
improving NAS performance. Conclusions are presented in Section 6. 

2. Analytical Framework 

The cost fimction of a firm is defined as the lowest cost at which it can produce a given 

set of outputs, Y , given the prices is pays for inputs, P . Equivalently, it represents the 

cost of acquiring the optimal set of inputs, X* , given the outputs and prices. Thus we 
have: 

COST,=P,-X\y„P,) = C{Y„P,) (1) 

where the subscript i denotes a particular firm (airline), and t identifies the time period. 
The cost function, like the production fiinction, is a way of depicting the technology 
available to the firm, i.e. its ability to transform inputs into outputs. Implicit in (1) is that 
all airlines have the same technology, an assumption that could be relaxed by adding 

airline subscripts to the cost and conditional demand (X*) functions. 

Equation (1) can be considered a long-run cost function because it assumes that all inputs 
have been adjusted to their optimal levels. Some inputs, particular capital inputs, cannot 
be varied instantaneously. A short-run cost function relaxes the assumption of optimal 
capital stock by treating capital as a quasi-fixed factor and removing capital costs from 
the dependent variable. This results in capital being an argument in the short-run cost 
function. Thus we have: 

SCOST,=P,.X\Y,,P,,K,) = SiY,,P,,K,) (2) 

where capital is excluded from the price and conditional demand vectors. 

It has long been recognized that costs depend upon the nature and quality of airline 
outputs as well as the quantity. For example, airlines have been shown to have economies 
of density, whereby the cost for a given total output increased with the size of the 
airline 's network. Several additional variables are included to capture such effects. These 

are incorporated into the vector Z„ . This yields a short run cost function of the form 

0{Yj,,Pii,Zj,,Ki,). This form, as well as the long-run version in which capital is not an 
argument, has been widely studied in the airline economics literature (Caves et al, 1985; 
Gillen, Oum and Tretheway, 1990; Windle, 1991; Encaoua, 1991; and Hansen and 
Kanafani, 1989), and serves as the point of departure for the present study. 

In this study we add one additional vector argument, N^ , which characterizes airline i 's 

operational experience in the NAS during time period t. In general, //,-, is based on 
variables such as average delay, delay variance, and the proportion of cancelled flights. It 
can be viewed as the performance of the NAS from the standpoint of an individual 

airline. This is not to suggest that A',, depends only on the performance of public aviation 
infrastructure; rather it derives from the interaction between that infrastructure and 
operational decisions taken by the airline. Both public and private investments in the 

NAS are primarily intended to change A',, for the better. 



Thus our analysis revolves around estimating the operating cost fiinction 
0{Yi,,Pu,Qi,,Kj,,Ni,) . The first four arguments are standard ones in the airline cost 
estimation literature. The last, which is the focus of our investigation, implies a 
relationship between NAS performance, measured at the airline level, and airline 
operating cost. In order to quantify that relationship, one must find a develop an 
A^„ vector which captures airline-level NAS performance in a compact, yet 
comprehensive, way. To this task we now turn. 

3. NAS Performance Measurement 

Our measures of NAS performance are derived from the operational experience of 
airlines using the NAS, as captured by such metrics as average delay, variability of delay, 
and flight cancellation rates. As noted previously, these measures do not only reflect the 
quality of service provided by the public aviation infrastructure, but also the airlines ' 
ability to plan and manage their operations. Both of these factors depend on exogenous 
events, particularly weather, as well as the competence (and perhaps luck) of service 
providers and users. Thus, when we refer to high or low performance levels, we are not 
affixing credit or blame to either the FAA or the airlines, but rather assessing operational 
outcomes in which both, along with a host of exogenous factors, played a role. 

We must quantify NAS performance by airline and quarter. To do so, results for 
thousands of flights must be summarized by a much smaller set of metrics. There is no 
uniquely valid way of doing this. One might, for example, base metrics on the flight, the 
flight complex, the day, or the airport-day. (To illustrate the last possibility, one could 
might categorize for a particular airline, airport, and day as Smooth", ttiildly irregular", 
or ftighly irregular", count the number of airport-days in each category, and use these as 
the performance metrics.) Here, we opted for the more conventional flight-based 
approach, reserving the others for subsequent work. 

Even while confining ourselves to flight-based metrics, there is a huge number that might 
be employed. To keep the analysis tractable, we employed a two step approach. In the 
first step, we evaluated seven metrics for each airiine and quarter in our data set. Next, 
v/e employed principal component analysis to collapse these metrics into a smaller 
number of factors, and calculated the factor scores for each airiine and quarter. These 

factor scores were used to compose the jV„ vector used in the subsequent cost 
estimation. 

The seven underiying metrics are defined in Table 1. The first two metrics pertain to 
delay, and are thus the most closely related to the conventional approach for measuring 
NAS performance. The third metric focuses on more extended delays and reflects the 
hypothesis that such delays may have qualitatively and quantitatively different impacts 
on costs. The next three metrics reflect variability in flight operations. The final metric 
reflects the incidence of conditions when operations become sufficiently irregular to 
result in flight cancellations. All of the metrics were evaluated by airline and quarter— for 
the 1 1 quarters extending from the winter of 1995 through the summer of 1997~using the 



Airline Service Quality Program (ASQP) data base, which presents scheduled and actual 
departure times for every domestic flight of the top 10 U.S. carriers. Thus our data set 
includes 110 observations. Since we employ a log-linear cost function specification, and 
all metrics were consistently positive, logarithms of these metrics are used in the 
subsequent analysis. 

As shown in Table 2, the seven performance metrics are highly intercorrelated. All 
correlations are greater than 0.4, and the majority are in excess of 0.6. This suggests the 
use of principal component analysis as a way of capturing most of the information 
contained in the seven performance metrics in a smaller number of variables. Principal 
component analysis identifies a set of factors—linear combinations of the original 
variables — which together account for as much of the total variation in the original data 
as possible. The factors are obtained by finding eigenvectors of the correlation matrix. 
The higher the eigenvalue, the greater the explanatory power of the associated 
eigenvector. By convention, each factor has zero mean and unit variance. By virtue of 
being eigenvectors, the factors are also mutually orthogonal. 

The results of the principal component analysis of the NAS performance data are 
summarized in Table 3. The first component has high, positive, loadings on all seven 
factors and accounts for 72 percent of the total variation. The second factor, which 
accounts for about half of the residual variation, has positive loadings on the variance 
metrics and negative loadings on delay and unreleliability. Thus airlines which score high 
on this factor tend to have unusually high delay variances combined with unusually low 
delay averages. The third factor explains 8 percent of the total variation, and has a high 
positive loading on flight cancellations and a negative loading on departure delay 
variance. Altogether, these factors explain about 94 percent of the variation in the total 
data set. 

In principal component analysis, the standard procedure is to determine the number of 
factors to be extracted from the data, and then rotate these factors so that factor loadings 
are close to either or ±1, in order to simplify their interpretation. In choosing the 
number of factors, one must make a judgment about when the additional variation 
explained by a factor is sufficient to justify retaining it. In the present case, we decided to 
confine our attention to no more than three factors, since none of the remaining ones 
accounted for more than 3 percent of the total variation. The choice between one, two, or 
three factors was more difficult. An oft-cited rule-of-thumb is to include only those 
factors which account for more than one Nth of the variation, where N is the number of 
variables in the data set. Applying this to the present case, we find that only one factor 
should be retained. On the other hand, this leaves out nearly 30 percent of the variation in 
the original data set, suggesting the possibility of adding a second or third factor. Rather 
than fixing on a single alternative, we chose to estimate cost models including one, two, 
and three NAS performance factors. 

Varimax factor rotation was then performed on the two factor and three factor 
representations. The results appear in Table 3. In the two factor case, the first factor 
correlates more hiahlv with the delav variables, including the average delavs. average 



delays over 15 minutes, and unreliability. The second factor has the highest loadings on 
the departure and arrival delay variances, the cancellation rate, and, like the first factor, 
average delays over 15 minutes. One might summarize this by terming the first factor 

tlelay", and the second factor irregularity". When three factors are used, the first one is 
virtually identical to that in the two factor case. The second factor is also quite similar, 
except that the loading on cancellation rate is considerably lower. The third factor has a 
very high loading on cancellation rate, along with some correlation with arrival delay 
variance and average delay over 15 minutes. The three factors might be described as 

tfelay", '^'ari ability", and disruption". A carrier with a high score on the first factor has 
flights that depart and arrive later (relative to schedule) than those of the average carrier. 
If the second factor score is high, than delays fluctuate more widely than average, while a 
high score on the third factor means that conditions in which flights must be cancelled are 
more prevalent than average. 

Figures 1 and 2 present average factor scores for the one-factor analysis, by airline and 
quarter respectively. Figure 1 reveals that, using the one-factor analysis, the two carriers 
experiencing the best NAS performance (i.e. with the lowest factor score) are USAir and 
Southwest, while Delta, United, and TWA experience the v/orst performance. Figure 2 
shows that the quarters with the worst NAS performance include the winters of 1996 and 
1997, along with the summer and fall of 1996. Good quarters include the springs and 
summers of 1995 and 1997. While there is some seasonal pattern in the data, it is not 
particularly strong, as evidenced by the fact that two of the three summer quarters are 
among the best while the third is among the worst. 

Figures 3 and 4 present airline and quarterly averages for the three factor analysis. These 
provide a more complete picture of NAS performance trends. We see that Southwest is 
the only carrier to be better-than-average for all three factors, while United is the only 
one to be below average for all three. A number of carriers feature performance far better 
than average for some factors and worse than average for others. For example. Northwest 
has relatively low delay (Factor 1), but high variability and disruption (Factors 2 and 3). 
In contrast. Delta has low disruption but high variability and delay. Because the factors 
are, by construction, orthogonal the lack of a consistent pattern in the airline factor scores 
is to be expected. 

From Figure 4, we see that just two quarters— the spring and summer of 1995— have 
better than average performance on all three dimensions, while two others— fall, 1996 
and winter, 1997— are consistently worse than average. We also see from Figure 4 that 
the horrific winter of 1996 was particularly bad from the standpoint of delay and 
disruption, but average from the standpoint of variability. A similar, but less pronounced 
pattern is seen in the winter of 1997, while in the winter 1995 only disruption was worse 
than average. Disruption is consistently less of a problem in the spring and summer 
quarters, as is delay except for 1996. Finally, there is some evidence of a secular trend to 
worse performance on the variability dimension: four of the first five months are above 
average in this respect, while each of the last six months is below average. 



4. Cost Model Specification and Estimation 

We now consider the relationship between the airline-level NAS performance factors 
derived in Section 3 and airline operating cost, using the cost function framework 
explained in Section 2. To do this, we use the performance factors to compose the NAS 

performance vector, Nj, , which in turn is used as an argument for the cost function. 

The airline cost estimation literature has evolved sophisticated techniques involving 
flexible ftinctional forms combined with simultaneous estimation of cost and input share 
equations. Here, we opt for a simpler approach— based on the Cobb-Douglas form-for 
several reasons. First, our data set is comparatively small, extending over just 1 1 quarters 
for which ASQP data were readily available at the time of our analysis. This makes it 
important to conserve degrees of freedom by using models requiring few parameters. 
Second, our aim is not to fully reveal airline cost structure, but simply to assess the 
impact of NAS performance on airline costs. Finally, we find that a simple Cobb-Douglas 
model fits the data extremely well, suggesting that more complex model would provide 
little 'i'alue-added." 

The Cobb-Douglas form leads to the log-linear model specification: 

\n{TOC, ) = ao + a, + 1 /3j ln(7,,, ) + X c;^ ln(F^,, ) + Yr( KZtu ) + x^ HK^, ) + 
J k e 



where TOCj, is operating expense for airline i in time period t; 

Yjit is the quantity of the output j for airline i in time period t; 

f'F^,, is the factor price for input k for airline i in time period t; 

Z^,y is the value of operating characteristic i for airline i in time period 

t; 
A^,, is working capital for airline i in time t; 

N^i, is the value for NAS performance factor m for airline i in time t; 

f „ is a stochastic error term. 

The specific variables included in the model are detailed in Table 4. Two outputs, 
revenue passenger miles, and ttther", are considered. The latter combines freight ton- 
miles, mail ton-miles and other miscellaneous outputs in a divisia index normalized so 
that this output is 1 for American Airlines in the first quarter of 1995. Three production 
factors, fuel, labor, and ftiaterials", are included. Fuel and labor prices are calculated 
using fuel expense per gallon and labor expense per employee respectively. The latter is 
somewhat imprecise because it does not take into account hours worked or employee 
classification (pilots versus flight attendants for example). As a proxy for ttiaterials" 
price, we use the producer price index (PPI), which varies by quarter but not by airline. 
The three operational characteristics are average load factor, the number of points served, 
and scheduled departures. These variables capture qualitative features of an airline 's 



output that are likely to influence cost. Our measure of airline capital stock is the sum of 
the airline 's net asset value, working capital, and accounts receivable, minus accounts 
payable. The capital stock variable is subject to some error because of the rather arbitrary 
depreciation rules used by airlines. With the exception of the PPI, all of these data are 
obtained from the airline balance sheet data published in the Department of 
Transportation 's Form 41 database. 

As previously noted, we employ NAS performance factor scores to define the //,-, vector. 
We estimate models in which this vector contains one, two, and three factor scores, 
employing the rotated factors. As a result of the rotation, the factors employed in the 
three models are all different from one another, as shown in Figure 3. By virtue of being 
factor scores, all have zero mean and unit variance. Also, because the factors are linear 
combinations of the logarithms of the seven original performance variables, they enter 
into the model in linear rather than log-linear form. 

The specification of the intercept term, a , is an important issue. As specified in (2), the 
model incorporates airline-specific intercepts, or airline fixed effects. Alternatively, one 
might assume a single intercept that applies to all airlines by eliminating cr,- from (2). 
When fixed effects are incorporated, they may absorb variation that is really due to other 
factors, particularly when analyzing short time series in which explanatory variables do 
not change very much for individual observations for the same airline. On the other hand, 
one could argue that individual airlines will tend to have higher or lower costs, all else 
equal, due to differences in productivity and other omitted variables. In the present 
context, we choose to include airline fixed effects because our focus is on the NAS 
performance variables. On the one hand, these variables do exhibit considerable intra- 
airiine variation, mitigating the absorption problem. On the other hand, consistent inter- 
airline differences in these variables may reflect differences in managerial competency 
that carry over into other areas, creating the possibility of spurious results if the dummies 
are excluded. We employ American Airlines as the baseline" carrier whose fixed effect 
is forced to zero. 

In order to efficiently estimate the model, it is desirable to account for the expected 
correlation between stochastic errors for observations pertaining to particular airiines. We 
do this by allowing first order correlation between observations for the same airiine. Thus 
we have: 

^ii = P£i,-\ + yu (3) 

where v,-, is the component of the error term which is independently and identically 
distributed. 

The model was estimated using the two-step Prais-Winsten (1954) method, in which p is 
estimated by performing regression on the OLS residuals, and the model is then re- 
estimated using the transformation v'u = V^ - pf,-,_, , where F„ is the vector of 
dependent and independent variables for airiine i and time period t, to eliminate the 
autocorrelation. To maintain degrees of freedom, the first observations in the time series 



are included, but multiplied by the factor -^l- p^ to maintain homoskedasticity. This 
estimation method bas been reported to be as efficient as full maximum likelihood 
estimation in simulation experiments (Johnson, 1984). 



5. Estimation Results 

Table 5 summarizes our estimation results for the cost models with one, two, and three 
NAS performance factors. All three models have very good fits, with R^ values in excess 
of 0.99. Coefficient estimates are of the expected sign, and most are significant at either 
the 5 percent or 10 percent level. The estimates are also, for the most part, quite 
consistent across the three models. As anticipated, the airline fixed effects absorb 
persistent inter-airline differences, reducing some of the other coefficients. To illustrate, 
consider the cost impact of Scaling up" an airline by increasing its outputs, departures, 
number of points served, and capital stock by the same proportion. According to the 
model a 1 percent scale-up would increase cost by P\+ fij -^Yx+Yz +/c percent. On the 
basis of the estimates this sum ranges from 0.79 to 0.88 percent, implying fairly strong 
returns to scale. On the other hand, there is a fairly strong positive correlation between 
the magnitude of an airline 's fixed effect and its scale of operation: Alaska, America 
West, and TWA have the smallest fixed effects, while American, Delta, and United have 
the largest. This suggests that some of the cost impact of scale-up has been shifted from 
the scale coefficients themselves to the airline fixed effects. 

Similarly, the estimate for the labor factor price coefficient is somewhat less than 
expected. This may also be caused by the fixed effects, or error-in-variables for the 
reasons explained in Section 4. Nonetheless, the estirnated cost function is approximately 
homogenous of degree 1 in factor prices {co^+co2+co-i='^) ^ predicted by economic 
theory, with the discrepancies well within the standard errors of the factor price 
coefficients. Error-in-variables and the presence of fixed effects can also explain the 
small and insignificant (though correctly signed) estimate of the capital stock variable. 

Turning now to the focus of our inquiry, we find that, in the one factor model, the impact 
of the NAS performance metric has the expected sign, but is statistically insignificant. In 
both the two and three factor models, however, one of the factors has the correct sign an 
is highly significant. In the two factor model, it is the second factor, which we termed 
frregularity" in the earlier discussion, which has the dominant effect. In the three factor 
model, the key factor is disruption. Recall that, in three factor model, the second and third 
factors, 'Variability" and disruption," are essentially a decomposition of the second 
factor, irregularity", in the two factor model. Thus the results are quite consistent, and 
reveal that it is the tiisruption" component of irregularity" that is the main cost driver. 
These effects are apparently lost in the one-factor model because the are subsumed in a 
single metric which also contains performance dimensions, such as average delay, that do 
not strongly influence cost. It is ironic that conventional investment analyses rely almost 
exclusively on these latter, seemingly unimportant, dimensions of NAS performance. 



To assess the magnitude of the link between cost and NAS performance implied by our 
results, recall that the factors are standardized variables, and thus have unit variance. 
Therefore, a one unit change in a factor corresponds to change by one standard deviation. 
From the estimates, we see that such a change in either the irregularity " factor in the 
two-factor model or the disruption "factor in the three-factor model will cause a change 
in operating cost of roughly 1.5 percent. Put another way, we estimate that, in either of 
these cases, a relatively good factor score of one standard deviation below the mean 
would results in costs about 3 percent lower than a relatively bad score of one standard 
deviation above the mean. 

In interpreting these results, it is important to remember that they reflect statistical rather 
than accounting relationships. Thus, in the three factor model, the strong impact of the 
disruption" factor, whose highest correlate is the flight cancellation rate, does not mean 
that canceling flights in and of itself is an important cost driver. Rather, the cancellation 
rate should be viewed as an indicator for the incidence of highly degraded operating 
conditions, in which there are many flights with high delays, heavily corrupted flight 
complexes, large numbers of stranded passengers, and so on. It is probably these 
conditions, rather than the specific act of canceling a flight, which generate the cost 
impact. 

6. Potential Benefits from Improved NAS Performance 

In this section, we employ the estimation results presented in Section 5 to estimate the 
potential gains, in terms of reduced airline operating cost, from improved NAS 
performance. The estimates we present are, in a very rough way, comparable to estimates 
of the cost of delay to U.S. airlines, such as those reported by Citrenbaum (1998), the 
FAA Airline Policy Office (1995), Odoni (1995), Geissinger (1989), and several others. 
Our estimates differ from these others in two important ways, however. First, they are not 
based on delay but on the broader concept of NAS performance. Second, they are based 
on cost comparisons involving a scenario in which performance is substantially 
improved, but delay is not eliminated. Thus, whereas the studies cited estimate the cost 
savings from the impossible feat of reducing delay to zero, here we estimated the savings 
from a conceivable, albeit dramatic, improvement in NAS performance. 

To ensure that the hypothetical improvement is realistic, we base it directly on the 
performance levels observed within our data set. Specifically, we calculate, for each of 
our 1 10 observations, the quantity X^-m^m/M as defined in (2), which represents the total 

m 

contribution (which may be positive or negative) of NAS performance to ln(rOC,-,). Let 

/ and / be the airline and time period for the obser\'ation, which we term the Reference 
observation", in which this contribution is the most negative. Then, for any other 
observation, we compute the cost savings that would result from changing the 

performance vector for that observation, yV,-, , to A',-,,.. Since we have about 100 
obsen-'ations, this procedure in effect considers a scenario in which NAS performance, 
measured at the airline, quarterly level, is consistently at the top 1 percent of what is 
presently experienced. 



We carried out this procedure for both the two-factor and three-factor models. In each 
case, the reference observation was found to be Southwest Airlines in the 3"* quarter of 
1995. Table 6 compares the faw" performance metrics for this observation with sample 
means over all 110 observations, revealing the magnitude of performance change being 
hypothesized. The reference cancellation rate and delay variances are more than 50 
percent below the corresponding sample means. For the other metrics, the differences are 
less pronounced although still considerable. On the whole, the comparison confirms that 
//,-.,. represents a marked improvement over present-day conditions, but not an 
unreachable ideal. 

Table 7 summarizes the cost savings from the improved performance scenario. Estimated 
annual operating cost savings from improving NAS performance are in the $1.5-$2 
billion range. The lowest estimate is $1.3 billion in 1995, based on the three-factor 
model, while the two-factor model applied to 1997 yields the highest estimate~$2.3 
billion. In general, savings estimated using the two-factor model are somewhat greater, as 
are those for the more recent years. This reflects the trend toward lower performance 
levels shown in Figure 4. The distribution of savings among airlines is naturally 
correlated with carrier size, with the largest airlines saving several hundred million per 
year, and the smallest ones about a tenth of that. Carrier savings also reflect their baseline 
performance levels, since those with poorer performance have more to gain. 

For the reasons explained previously, these estimates are only roughly comparable to 
previously published ones of the cost of delay. Nonetheless, the latter offer usefiil 
benchmarks. The most recent published estimate, due to Citrenbaum and Juliano (1998), 
places the total direct operating cost of delay to air carrier and air taxi operators at $0.8 
billion in 1 996. However this estimate is derived solely from comparisons between actual 
and scheduled gate-to-gate time, and thus does not consider costs of departure delays nor 
the phenomenon of schedule padding. Earlier FAA estimates (Aviation Policy Office, 
1 995 are based on arrival delays instead of gate-to-gate delays, and yield annual figures 
of $2.5 billion, in current year dollars, throughout the early 1990s. Geisinger (1989), 
disaggregates delay by phase of flight and applies different cost factors for each phase, 
and obtained a cost of SI. 8 billion in 1986 (using the ATA composite index, this equates 
to $2.5 billion in 1997). Odoni (1995), places the cost of delay, non-optimal flight 
trajectories, and flight cancellations to airlines in the $2-4 billion range in 1993. Our 
figures of $1.5-$2.3 billion clearly fall within the range of these estimates. However it 
must be reiterated, unlike the other estimates, ours are based on a comparison with a 
realistic performance scenario rather than a perfect one. In that respect, our results 
suggest a greater potential saving from attainable performance improvements than the 
prior studies. 



7. Conclusions 

Our results support the view, suggested by several earlier studies, that improvements in 
the performance of the NAS can generate billions of dollars in annual cost savings. 
Unlike previous work, however, the estimates presented here derive from observed 



12 



covariation between airline expenditures and NAS performance levels. As a result, they 
do not rest on the strong and implausible assumptions required to calculate costs from 
quantities of delay, nor even on the assumption that delay is the critical cost driver. It is 
reassuring that such a fundamentally different methodology yields potential savings of a 
comparable magnitude. 

Despite this agreement as to the bottom line" our study presents a qualitatively different 
view of the link between NAS performance and airline cost. Of the performance metrics 
considered, we find quantities of delay to be among the least important. Instead, we find 
the critical cost drivers to be the levels of irregularity and disruption in the system. If we 
had to choose a single metric to track this dimension, it would be the flight cancellation 
rate rather than the average delay per flight. This may have significant implications for 
how NAS investments should be prioritized. In general, investments that increase the 
1'obustness" of the system by preventing all hell from breaking loose" appear to be 
more promising than those leading to incremental delay reductions in a broader range of 
conditions. 

Methodologically, this study points to the role of statistical cost modeling as a means of 
translating the emerging, multidimensional, view of NAS performance into improved 
capability for investment analysis. Any dimension of NAS performance that can be 
measured at the airline level can, in principal, be related to airline cost using the methods 
set forth here. The only practical limitation is that the impact be strong enough to be 
detectable through the statistical noise. As data accumulates, our detection capability wdll 
improve. 

As previously noted, there are other approaches to representing NAS performance that 
may more aptly capture cost impacts. One approach would be to categorize days, or 
airport days, in terms of their regularity and base performance metrics on the number of 
days in each category. Another would be to categorize total delay minutes according to 
type of flight, phase of flight, duration, and other factors and then develop metrics that 
summarize how delay is distributed across these categories. Other investigative 
approaches, including structured questioning of airline decision-makers and detailed 
simulations of airline operations, may also be of value. Such work may ultimately enable 
analyses of public and private investments in aviation infrastructure which capture their 
true benefits. 



13 



References 

Allen, David, Aslaug Haraldsdottir, Robert Lawler, Kathleen Pirotte, and Robert Schwab, 
The Economic Evaluation of CNS/ATM Transition, "Boeing Commercial Airplane 
Group, 1998, 
http://ww\v.boeing.com/commercial/caft/reference/documents/caft paper.pdf 

Alcabin, Monica, 'Airline Metric Concepts for Evaluations Air Traffic Service 
Performance," CNS/ATM Focused Team, Air Traffic Services Performance Focus 
Group, 1999, 
http://www.boeing.com/commercial/caft/cwg/ats perf/ATSP Febl Final.pdf 

ATS Data Link Focus Group, Data Link Investment Analysis, " CNS/ATM Focused 
Team, 1 999, http://www.boeing.com/commercial/caft/cwg/ats dl/tocpaper.pdf 

Caves, D.W. , Christensen, L.R. and M. W. Tretheway, Economies of density versus 
Economies of Scale: Why Trunk and Local Service Airlines Differ", Rand Joumal of 
Economics,15 (1984), 471-489 

Citrenbaum, Daniel, and Robert Juliano, 'A Simplified Approach to Baselining Delays 
and Delay Costs for the National Airspace System, "Federal Aviation Administration, 
Operations Research and Analysis Branch, Preliminary Report 12, 1998. 

Encaoua, D. Liberalizing European Airlines: Cost and factor Productivity Evidence" 
International Joumal of Industrial Organization (1991),9, 109-124 

Federal Aviation Administration, Office of Aviation Policy and Plans, Economic 
Analysis of Investment and Regulatory Decisions-Revised Guide (1998), Appendix A, 
Documents Requiring Economic Analysis, http://api.hq.faa.gov/apo3/appenda.PDF 

Federal Aviation Administration, Office of Aviation Policy and Plans, Total Cost for Air 
Carrier Delay for the Years 1987-1994," 1995, http://api.hq.faa.gov/dcosl995.htm 

Federal Aviation Administration, Office of System Architecture and Investment 
Analysis, NAS Architecture 4.0, 1999, http://www.faa.gov/nasarchitecture/version4.htm 

Geisinger, Kenneth, 'Airline Delay: 1976-1 996, "Federal Aviation Administration, 
Office of Aviation Policy and Plans, 1988 

Gillen, David, Tae Oum and Michael Tretheway "The Cost Structure of the Canadian 
Airline Industry", Vol. 24, No. 1 Journal o'f Transport Economics and Policy . (January, 
1990)9-34 

Johnson, John, Econometric Methods, Third Edition, McGraw-Hill, 1984. 



14 



Kostiuk, Peter, Eric Gaier, and Dou Long, The Economic Impacts of Air Traffic 
Congestion, "Logistics Management Institute, 1998, 
http://www.boeing.com/commercial/caft/reference/documents/lmi_econ.pdf 

Odoni, Amadeo, Research Directions for Improving Air Traffic Management 
Efficiency, "Argo Research Corporation, 1995 

Prais, S., and C. Winsten, Trend Estimates and Serial Correlation, Cowles Commission 
Chicago, IL, 1954 

Windle, R.J., 'The World 's Airlines: A Cost and Productivity Comparison", Journal of 
Transport Economics and Policy, 25 (1) 31-49 (1991) 



15 



Table 1. Performance Metric Definitions 



Variable (in Log form) 



Average Arrival Delay 



Average Departure Delay 



Average > 1 5 min Arrival Delay 



Arrival Delay Veiriance 



Departure Delay Variance 



Unreliability 



Cancellation Rate 



Deflnition 



Difference between scheduled and actual 
arrival time, averaged over all flights. 



Difference between scheduled and actual 
departure time, averaged over all flights. 



Sum of all arrival delays in excess of 15 
minutes, divided by total number of flights. 



Variance of the difference between 
scheduled and actual arrival time. 



Variance of the difference between 
scheduled and actual departure time. 



Proportion of flights with an arrival delay 
over 15 minutes. 



Proportion of flights cancelled. 



Table 2. Results of Principal Component Analysis 



Variable (in Log form) 


Factor 1 


Factor 2 


Factor 3 


Average Arrival Delay 


0.85468 


-0.46316 


-0.05345 


Average Departure Delay 


0.86167 


-0.31608 


0.07645 


Average >15 min Arrival Delay 


0.98444 


0.02062 


-0.03112 


Arrival Delay Variance 


0.81784 


0.51105 


-0.08456 


Departure Delay Variance 


0.77662 


0.38233 


-0.42532 


Unreliability 


0.91327 


-0.31949 


-0.02627 


Cancellation Rate 


0.70281 


0.31987 


0.61739 


Proportion of Variance Explained 


0.7203 


0.1324 


0.0828 


Cumulative Proportion 


0.7203 


0.8527 


0.9355 



16 



o 



O 

O 
b 
1/1 



O 

H 
U 

< 
to 

< 
H 
O 

Pi 

<: 







^ 


, , 


^^ 


«n 


a\ 


(N 


, 


0\ 


rs 


W-) 






c 


■* 


(N 


m 


m 


m 


m 


m 


0\ 


m 




M 


.2 


■^ 


r- 


o 


>r» 


lO 


r-~ 


in 


00 


r*1 




o 


00 


(—4 


t-- 


rt 


ON 


,-^ 


(—1 


t— 4 


On 




■»-» 


<N 


— " 


m 


rl- 


o 


tN 


ON 


d 


d 






5 


d 


d 


d 


d 


d 


d 


d 








R 


e/) 






















&H 

























s 


o 


r-- 


,_ 


o 


OS 


, 


r~- 


w-i 


(N 


hree 
ctors 




>-. 


t— 1 


m 


kO 


rj- 


o 


ON 


»n 




NO 


fS 


"t-* 


r<^ 


00 


o 


(N 


>o 


>n 


«/^ 


o 


rl- 


I. 


•— I 


'3- 


-— < 


00 


(N 


o 


CN) 


in 


m 


r^ 


o 


IS 


(N 


(N 


>r> 


OO 


On 


ro 


(N 


d 


d 


^^ 


u 

es 
(I. 


.2 

CO 


d 


d 


d 


d 


d 


d 


d 










r^ 


o 


m 


o 


r- 


■* 


in 


r«- 


t^ 




^ 


5 


<n 


w-i 


•j^ 


VO 


(N 


O 


m 


■* 


•^ 




^, 


>, 


r—* 


t— < 


•rt 


0\ 


ON 


in 


(N 


Tf 


Tf 




o 


CO 


rl- 


■* 


O 


■<a- 


1—1 


oo 


t^ 


Ti- 


■^ 






u 


00 


0\ 


r-; 


(N 


m 


oo 


(N 


d 


d 




Q 


d 


d 


d 


d 


d 


d 


d 








Ce, 


























J 


rl- 


00 


o. 


t~- 


c~- 


m 


Tj- 


rf 


r- 






V. 


OS 


00 


C\ 


w-» 


oo 


ro 


ON 




(N 




<s 


*-• 


C^ 


vo 


o 


On 


NO 


rs 


o 


o 


lO 




o 


'C 


''J- 


m 


oo 


fS 


o 


oo 


^^ 


TT 


oo 




J2 


m 


(N 


VO 


Os 


oo 


r-1 


t^ 


d 


d 






3 

00 


d 


d 


d 


d 


d 


d 


d 








R 


u 




















« " 


b 


fc 




















? o 




•H 




















^ -^ 






































































Cb 






oo 


o 


o 


CN 


00 


o 


NO 


■* 


•* 




^H 


5 


■^ 


oo 


fN 


t- 


Ti- 


oo 


m 








t^ 


>^ 


00 


(N 




NO 


ro 


oo 




^r\ 


m 




o 


^ 


■<t 


■<a- 


»— « 


m 


^— 


oo 


o 


-* 


■* 




u 

« 


"u 


00 


ON 


(^ 


fN 


m 


00 


m 


d 


d 




P 


d 


d 


d 


d 


d 


d 


d 








U. 




























00 


r~- 


■* 


Tj- 


CN 


r- 


^^ 


(^ 


m 




^H 




VO 


VO 


■^ 


00 


NO 


(N 


00 


o 


o 


« 2 


o 
u 

b 




■^ 


.— • 


■^ 


r^ 


NO 


m 


<N 


<N 


fS 




1/^ 


VO 


00 


p— • 


r^ 


1-^ 


O 


t~~ 


r~- 


O « 




oo 
d 


oo 
d 


ON 

d 


oo 
d 


d 


On 

d 


d 


d 


d 






















•o 






^^ 








C3 










4) 

C 






£ 








O 










]3 






en 

e 




>^ 




Q 
> 




1) 

.5 






X 

u 


c 




^ 




C3 


Q 




f— 


'u 






o 




o 








4) 
Q 

> 


■d 

CU 


< 

C 

E 


> 


C3 
> 

a 
o 


>, 


c 
_o 


.2 
ea 

> 

O 


■■5 
o 

S 

(X 

4> 




" 




< 


r^ 


a" 




y 


~ 


o 


> 




^' 




o 


o 


i) 




L. 


^ 


R 


■^- 




^ 




OD 


jp 


=p 


— 


^ 


« 


^ 


■■5 


C5 








« 






c^ 


1. 


•"* 


o 


o 


3 










'1^ 

o 




> 


cl 


H 


o 

c 


2 


E 








p- 


> 


> 


■^ 


o 


C 


n 


3 








< 


< 


< 


< 


'■N 


3 


U 


CU 


u 



Table 4. Cost Estimation Variable Descriptions 



Variable 


Definition 


QUARTER 


Quarter of year (l=Winter, 2=Spring, etc.) 


TIND 


Time counter (1 for IQ 95, 2 for 2Q 95, •••, 1 1 for 3Q 97) 


ALF 


Average load factor (revenue passenger miles/revenue seat 
miles) 


IDO 


Index of output other than passenger miles (cargo, freight, 
etc). Normalized to American Airlines in IQ, 1995. 


TOC 


Total operating cost for quarter ($) 


RPMS 


Revenue passenger miles (000) 


WAV 


Total labor expense per employee ($) 


WFUEL 


Fuel expense per gallon ($) 


WMAT 


Produce price index (proxy for price for materials and 
services) 


WK 


Working capital ($) 


SDEP 


Number of scheduled flights 


BASE 


Number of points served 


CARRIER 


Carrier code 


YY 


Year 



p 

O 

I— < 

S 

H 

en 

Z 

o 

H 
U 



H 

CO 

O 
< 







« 


1 


«P 


)P 


.. 


^ 


•^ 


VO 


w-v 


Q 


> 


■^ 


IN 


•» 


Wl 





OS 


^ 


■*s 


r*> 




j 


1 


r^ 


f*\ 


5 










£! 




<N 


r< 


<N 


> 


<o 


M 





so 


so 


so 


■* 


<a 




OS 


tri 


>o 


<n 


so 






1 


vc 











^ 


<a 




>o 


>o 


OO 


> 


*^ 


OS 


»n 


> 


Os 


<N 


«0 


*^ 


!n 


SO 


so 





> 










OS 










R 


OS 




V 


<s 


d 


s 


V 


— 


d 


N 


fn 


•"S 


"n 


<<> 


•n 


— 


— 


— 


Pi 


— 











— 


'n 






1 


— 


























































, 


o 


j 


Tf 


t- 


o 


o 


»ri 


r-i 


Tf 


Tf 


r^ 


>n 


00 


pn 


»n 


TT 


TT 


OS 


fS 


r~ 


j 


I 


! 


m 


VO 


^* 






e 


u 


o 




<N 


g 


s 


o 


§ 


s 


— 


r^ 





i»i 








3 


00 


r; 


^ 


§ 


s 






1 


8 


8 


i 






u 


— 




d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 








d 


d 


d 






,« 


TS 


























































k. 


on 


























































9> 


o 


1 


fn 


Wl 


o 


(S 


•<r 


c-» 


00 


*— 





^m 


00 


^ 


Os 


00 


00 


so 


fS 


^. 


J 


1 


1 


rl 


f-* 


r^ 


00 




£ 




oo 




r^ 


r* 


i3 


•T 


v-i 


s 


in 


m 


OS 


■V 





so 


Os 


s 


00 


in 





m 











8 




00 


H 


« 


•«r 




«n 


^- 


o 


•>r 


n 





m 


(S 


•9 




M 




(N 




<s 

















OS 


NO 




E 


<> 




o 


9 


9 


9 


9 


9 


9 


d 


d 

1 


d 


d 


d 


d 


d 


9 


d 


d 


9 








9 


d 


d 


On 






1 






























































•'1 


1 


« 


r 


o 


<o 


>o 


00 


OS 


Os 


<N 


Os 


t>. 


"s^ 


»*^ 


fS 


r4 


^ 


w^ 


fo 


1 


>n 


'n 


1 














^ 




> 


r*» 




» 


5> 




so 


>n 


so 


». 


Os 


>n 


'n 


»n 


ro 


S 


«o 


^ 


■ 




R 


1 












4-* 


Q 




t^ 


^ 


OS 


> 


<N 


s£) 


■o- 


> 


Os 


5s 


SO 


> 


tsj 


■<r 


00 





i^ 


<N 




q 


<v 












■a 
o 


s 

1 


o\ 




V 


<S 


d 


V 


>■ 




d 


iS 


"*i 


fn 


>n 


so 


"n 








N 








fn 












z 


, 


C\ 


1 


«n 


r~ 


~m 


w 


-o 


n 


so 


r- 


•^ 


t~- 


OS 


n 


»n 


^ 


^ 


,„ 


^ 


r* 


• 


en 


»n 


< 


t 


, 






i. 


v-i 




«s 


)S 


iS 


o 


00 


n 




m 


r- 


r-i 




^ 


pn 


00 


r^ 


»n 


00 


<N 


1 





8 


1 


t 








b 




o 




"^ 


o 


o 


— 


o 


o 


— 


















f^ 



























o 


— 




o 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 






















■o 


























































u 


«^ 


1 


Tf 


Oi 


lo 


m 


<s 


•«r 


•sa- 


so 


Tf 


•ST 


r~- 


tn 


(N 


r~ 


OS 


tN 


OS 


fN 


] 


rn 


VO 


! 


1 


, 


OG 


1^ 




00 




OS 


i^ 


'd 


w^ 


VO 


«/^ 


in 


m 


OS 


(N 





p^ 





tn 




SO 


OS 


m 









I 










•! 


M-i 




Tl 


— 


o 


•«r 


n 


o 





fo 


<N 


•<f 




fS 


fN 


>o 


^ 























Os 


00 




_E 


o\ 




d 


d 


9 


9 


9 


9 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


9 
















o5 




































































*^ 


1 


ee 


Q 


00 


C 


a 


M 





«s 


'^ 


<a 


*^ 


D» 


00 


t~ 








o< 


i~- 


r^ 


1 


1 


, 


, 


i 










0\ 




*^ 


*N 


m 


<s 


>o 


00 


w-» 


Oi 


Oi 


OS 


Q 


»v 


so 





t~ 


so 


QO 


•n 


Os 




( 






1 








^ 


<N 




>o 


Q 


r- 


"^ 


00 


00 


m 


■» 


«o 


Os 





(•^ 


^ 


eo 


00 


>n 







■«r 


















a 


oJ 




<n 


"5 


d 


wi 


V 




d 


<n" 


"i 


<n' 


«S 


so 


>^ 






d 


<n" 




















Z 


, 


VO 


1 


o\ 


o\ 


"o 


s 


oo 


•fl- 


-i- 


r~ 


t~ 


•w 





•<3- 


i~ 


m 


f^ 





OS 


00 


m 


1 


1 


1 


! 


1 






k 


52 




<N 


S 


so 


o 


oo 


m 


rM 


■» 


r~ 


■V 


CN 


s 


m 


OS 


00 


so 


00 


CN 









1 


I 








k. 


ts3 


o 




— 


o 


o 




o 


o 

















n 





























o 


«— 




d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 
















c 


2 
^ 


























































u 


i~ 


1 


■fl- 


2 


00 


— 


o 


^ 


t~ 


so 


^ 


_ 


(N 


00 


(N 





.. 


OS 


in 


ON 


sn 


! 


1 


1 


1 




ZT" 


f«. 


o 




00 




<N 


o 


s 


wi 


m 


'S 


^ 


so 





r^ 





Os 


OS 




•«■ 


00 


OS 


rj 





1 


• 


1 


t 


• 


00 




« 


o 




r^ 


fS 


o 


w*t 


^ 


fn 


1-1 


■9 




cs 




?^ 


l*S 























0\ 







E 


o 




9 


d 


d 


d 


d 


9 





d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 


d 












o> 


m 




1 




























































k. 




























































4i 














































^^ 




/— V 


/— ^ 




CJ, 


















































fi 














B 










































^ 




^* 




?r 


?r 




^ 




a 










































w — 


^^ 


K 


^^ 


""^ 


•»— ' 






Lb 

a. 












































r-i 


h 


3 


i> 


.0 






















































n 


























































_C3 


CO 


\ 




0. 

1 








o 










































> 







*- 







c 




8 




W1 




















^-^ 






































< 




^ 


^ 


^^ 


^^ 


-— ^ 


^^ 


^^ 


,-^ 


g 


u 




















tJ 


u 


U 


u 


td 


tj 




z 






^ 




>- 


?; 


^ 


^ 


^ 


^ 




S 




















u 





U 


u 





u 




g 




« 




ti 


(J" 


a" 


b" 


Si 


l? 


S^ 


G 


a 




















z 


Z 


z 


z 


2 


z 




r- 




^ 




'*-^ 








3 


^-^ 




^— ' 


■^^ 




















< 


< 


< 


< 


< 


< 


■• 


< 

-3 




« 




o 


Z 


o 


o 


— 


H 


T3 


— 




5 






^ 














2 


2 


s 


2 


2 


2 


C^ 




> 


? 


ID 












'J 




^ 




a: 


< 


2 


^" 










ei 






ci 



Ci 





p 


2 









1- 


■3 
< 


■3 


■3 
_X 


■3 

3 




X 


•3 

1 


X 


■3 

X 

2 
> 


"3 
•-> 
X 


=5: 

2 


^^ 


5 


^ 

^ 


K 


5 


Z 
p 




E 

^ 


c 
< 


■J-. 


c/: 
< 


< 


in 
< 


CO 

p 




u 

p 

5 




— 


< 


< 


G 


r^ 


— 


• 




— 


2 


5 




^ 


< 


5 




2 


2 


2 


Z 


Z 


2 


5 


< 






o 



< 
z. 

u 

C/2 




















He 

O 

•c 


c 
6 


C 

'e 

1—1 


'e 


s 

C 

E 


C 

E 


0\ 

o 


oo 
o 
o 




M 


r^ 


VO 


00 


o 


S 




d 




U 


u 


^ 


Tf 


^ 


m 


eN 








2; 


w 


















1 


w 

C 
n 


















p 


s 


















§ 


u 
£ 


















u 


u 


















Cm 






















■B 
« 

> 


















O 


£ 


















« 


c 
E 


















^H 




















Q 

z 
< 

z. 




c 


c 
E 


c 
E 


c 
E 


c 
E 


d 


o 

O 




< 




fS 


lO 


<N 


o 


M- 




d 




u 




od 


00 


c^ 


00 


r~ 








s 


e 
c: 

u 


















U 
hJ 




















< 


E 


















C/2 


















.2 


en 




















U 


















s 


P 


















> 


h:i 


















< 


















QO 


> 


















u 

^ 


WD 




















U 


















^ 


h-4 








>» 










i 


H 








_« 










u 

M 


Cd 








(U 










o 


u 
z 

< 


u 
> 


* 
♦ 

a 
u 

Q 

"a 
> 


* 

* 

Q 
u 

l-H 


1 

< 

c 
E 


o 

c 
2 

> 


o 
o 

c 

2 
c 

> 




u 




o 




C 
;- 


<u 


in 


>-> 




^^ 


c 


1> ^ 






< 


2| 


I 




2i 

3 


£ 


o 

*-< 


i a 


c^ 






CO 


C3 


« 


2 


O 


< o 






u. 


Lh 


k- 


> 


C3 


u 


O 


s; u 


^ 




> 


> 


«. 


E 




S 


c 

C3 


M 


bJ 
^ 




< 


< 


< 


< 


=^ 


D 


J ■ 


5 E 


^»A 




O c= 


^ 


^ ^ 








■S J3 




e/i '^ 




n < 


















C 


a » 



o 





OO(N(NOr->V0CN— <cs 


m 






v£>r<imr-r<^-^CNfN(Nm 


m 






Tj- .-. -^ (V) r-i CN 


r- 






fe^ 






•K 








t-» 








OS 








0\ 








1-H 








_ 








o 








^^rt—i— .^OOi^OOO 


o 




s 


r^(N<oCT\rfr>- — (/n<Nr^ 


rr 




ro ^- _ (-V) ^« r<^ (N 


t^ 




u 


c>^ 






o ^ 








r: ff\ 








Si On 








« 1-H 








•^ 








<u 








V 








k 








J3 








H 


— mmiri\oio(Nu->or~- 


r- 






o-Hra-mcNmoovnTj-cN 


lO 






ro „ ^ (VI (N — 


r<-i 






v» 






IT) 








C^ 








Cn 


















O'—oomt^^— 'ooioooo 


r<-i 






omor-'<a-ooo^om 


«n 






^ (N m (N — m (N 


<N 




■k 


<A 


<N 




r~ 








C\ 








0\ 








1-H 
















"3 








-o 


Otnoioo<N\ooomo\ 


00 




o 


r--<NONmTro-^oom 


Vi 




^ 


■^ r^ m m '-< Tf (N 

6^ 


rs 




u \o 






t~ 


22; 






OS 
CTv 


a ^ 






w. 


1 






O 


o 






k. 
o 




Tj-Ttr~oooT}-mTro\(N 


\o 


i 

CT 




O — VOt-^m^OsOsOCN 


r-- 






•^ — CS CN (N — 


\o 






6^ 




E: 


Ul 






j= 


CS 






«rf 


ON 






i2 


^i^ 






c 
o 

■o 
o 

CO 








C3 








X) 




*J 




•o 




CO 




u 




American 

Alaska 

Continental 

Delta 

America We 

Northwest 

TWA 

United 

USAir 

Southwest 




C5 

E 


11 

< 


o 


to 
o 

3 
C 




H 


< 



(N 



0) 

c 



■a 



lA 

"5 

c 
< 

u 
O 

o 



c 

o 

o 



w 



o 

re 
u. 
o 



0) 

> 
< 



0) 

u 
3 
O) 

iZ 







eJoosJojOBj a6BJ3Av 



.< {'■■ 



u 

■C 
re 

3 

a 

X) 



u 
re 
u. 

0) 

c 
O 



o 
u 
» 

o 
u 

re 
u. 
a> 

O) 

2 

lU 

> 
< 

s 

3 
S) 




GJU 



in 
en 

■> 
o 



I en 
Dcnl 

- 3 
< 



If) 



o 
in 



in 

Go 
— 0) . 



o 
o 



o 



O 
O 






■.-:,<» 












m 



^ 









c 
o 
S 

8) 



4) 

r 

n 

3 

o 



(N 



.'felF 



o 



o 
o 



o 
in 



3J03SJ0)3Bd sBejaAV 



I- 


(N 


" 








1 o 


o 


n 








o 


(> 


u 


ra 


TO 


ra i 



o 

•c 
< 

Si 



(A 

>< 



o 



o 
u 
w 



ffi 

O) 

2 

> 
< 

0) 

3 
O) 
11. 



□ D D 



o 
in 







ejoosJopej aBejBAV 



It- 


CM 


CO 1 


' i- 






: O 


O 


o 1 








o 


U 


() ' 


1 TO 


ra 


(D ' 


LL 


LL 


U. , 


jD 


■ 


u\ 



CO 

3 

o 



u 

u. 



o 
m 



o 

CO 
UL 

a 
oi 

C 
o 

> 
< 

ti 

b. 
3 




c 
o 

S 

•o 



O) 

c 

CO 

3 

O 



in 



ajoDs JOjOEj a6BJ3Av 



Relative Efficiency of European Airports 



Eric Pels', Peter Nijkamp*, Piet Rietveld* 

Free University Amsterdam, Department of Regional Economics, 

de Boelelaan 1105, 1081 HV Amsterdam, Netherlands. 

Email: apels@econ.vu.nl. 



Abstract 

Using data envelopment analysis, efficiency ratios for European airports are determined. It appears 
thatmost airports are operating under increasing returns to scale. This is also reflected in the most 
productive scale size determined for the airports. 

Keywords: airports, most productive scale size, data envelopment analysis 



* Also affiliated to the Tinbergen Institute, Keizersgracht 482, 1017 EG Amsterdam. 



1 Introduction 

Recently, a number studies have been published on the measurement of airport performance; see e.g. 
Gillen and Lall (1997, 1998) for a non-parametric approach using data envelopment analysis, Hooper 
and Hensher (1997) for an analysis using total factor productivity and Tolofari et al. (1990) for an 
estimation of translog cost functions for British airports. This interest is caused by Uvo circumstances: 
(i) the deregulation of the aviation market has stimulated the development of reliable performance 
measurements, since airlines operate in a highly competitive market and cannot pass the higher 
operating costs at inefficient airports onto the passengers (see Gillen and Lall, (1997), and (ii) the 
growth in the number of passengers raises the question whether this growth should be accommodated 
at existing airports or at new airports. This issue requires insight into the operating characteristics and 
performance of airports. 

The estimation of cost functions allows for the testing of several hypotheses concerning 
economies of scale and the technology concerned (see e.g. Tolofari et. al. (1990)). As input prices are 
rather volatile over time, estimation of a longer term cost function becomes rather difficult. Moreover, 
input prices, if available, can differ significantly (in the way they are collected and reported) in space. 
Data envelopment analysis (DEA) does neither require knowledge on the input prices nor assumptions 
concerning the production technology or on the behavior of actors (e.g. cost minimization). The only 
assumption is that the production possibility set is convex. Using DEA one can get either input or 
output oriented efficiency measures for decision-making units (dmu-s)'. DEA also allows for a 
determination of the most productive scale size (mpss). At mpss, the average productivity is 
maximized; it represents the maximum productivity for any given input-output combination. Using the 
same technique it is also possible to determine whether a dmu operates under increasing, constant or 
decreasing returns to scale. 

The DEA approach mentioned above provides a "measurement" of inefficiency (the "Farrell 
approach") rather than an "explanation" of inefficiency (the "Leibeivstein approach") (Button and 
Weyman- Jones, 1994). An airport can be labeled as inefficient for different reasons. First, there are 
"indivisibilities". An expansion of the runway system will, in most cases, automatically create an over 
-capacity, since the length of a (new) runway is mainly determined by the landing (or take off) weight 
and speed of the aircraft using that mnway. It may be necessary to construct a new runway, but due to 
technical (and safety) requirements, it may not be possible to make its capacity fit the expected 
(additional) demand. To a lesser extent the same holds true for terminals. Second, there are 
government regulations (e.g. limits to the hours of operations, noise contours) and limiting physical 
circumstances (e.g. fog and wind) under which the airports operate. For example, Amsterdam Airport 
Schiphol covers a (relatively) large area, also because of the noise contours imposed by the authorities. 
Schiphol has 4 runways in use, which, due to various weather conditions and strict regulations, have 
only a limited use. These two causes of inefficiency do clearly not fall under the control of the airport 



' By using the term dmu Chames et al. (197S) emphasize that their interest lies in the decisions made 
by non-profit organizations rather than the (in theor>') profit maximizing fuTHi. ; 



management, but on the other hand may not fully explain the inefficiency; airports can also be 
inefficient because the operators do not have an incentive to work as effectively as they could, e.g. if 
airports are government controlled. Although the DEA approach used does not require any assumption 
on the behavior of the actors involved, it is possible that, next to the (pure technical) inefficiencies 
described above, X-inefficiency also is important^. In this paper we are primarily concerned with the 
measurement of inefficiency. Gillen and Lall (1997) first measure inefficiency using DEA, and then 
explain inefficiency in a Tobit-analysis. We do not have sufficient data to explain any differences in 
efficiency in this way. 

Inefficiency can also be measured using the stochastic production frontier method. This is a 
parametric method, and assumptions on the production technology are necessary. Inefficiency can be 
measured and explained simultaneously, in contrast to an ad hoc explanation of the DEA inefficiency 
measures. However, in contrast to the DEA, specific assumptions concerning the production 
technology are necessary. 

The purpose of this paper is to measure the relative inefficiency of European airports and to 
indicate, given the prevailing input combinations, the maximum productive scale size for each airport. 
This scale size is the optimal configuration at the current input mix (i.e. if an aiiport were to change its 
technology, the maximum productive scale size would also change). 

The outline of the paper is as follows. In section 2 the concept of mpss will be described. 
Section 3 contains a short description of DEA and the empirical results are presented in section 4. In 
Section 5 the inefficiency measures from section 4 are compared to the inefficiency measures firom a 
stochastic production frontier analysis. Section 5 offers concluding remarks. 



Most Productive Scale Size 



In this paragraph we provide a concise description of the concept of most productive scale size. For a 
more elaborate exposition the interested reader is referred to e.g. Banker and Thrall (1992) and Banker 
(1984). Using a minimum cost mix of inputs one can determine an optimal scale of a firm. This 
optimal scale depends on the prevailing input and output prices. Hence knowledge of the prevailing 
prices is required. Moreover, prices can be more volatile than the technology used. When one is 
interested in the performance of a decision-making unit over a longer period (or when prices are not 
available), the (pure) technical relation between inputs and outputs becomes interesting. One can look 
at returns to scale, and, associated with that, at the most productive scale size (mpss) at a given input- 
output mix'. The mpss for a given input-output mix is "the scale size at which the outputs produced 
"per unit" of input is maximized" (Banker, 1984). Thus, the idea of mpss is related to average 
productivities. If a dmu is operating under increasing returns to scale, it can increase the output "per 
unit of input" by increasing its scale. If decreasing returns are prevailing, it can increase the output "per 



' Note that "regulators" do not necessarily have an incentive to reach a social optimum. Hence, 
regulations can also be a cause of X-inefficiencv. 



unit of input" by decreasing its scale. It follows that at mpss, a dmu operates under constant returns to 
scale (see Banker (1984), for more details). This is illustrated in figure 1. 



Figure 1 



Production possibility set and (in) efficient points of production 




I=(x,Xo,y,Yo); I=M,A,B,C . 
T={(X=xXo,Y=yYo)IX>0 can produce Y^O} 

At M average productivity (ON/MN) is maximized (i.e. M represents a mpss). Furthermore, 
(dyM/dxM)(xM/yM) = 1; the dmu at M operates at constant returns to scale. The efficiency of a 
production possibility A can be evaluated against M by taking the ratio h^ = (OD/DA)/(ON/MN) = 
{y/J''^/^/{ yJ^sd- This inefficiency measure captures both inefficiency due to technical inefficiency at 
the given scale {{yAf^Ayiy^^^i)) and inefficiency due to a divergence from mpss {{yB^x.^yiyJ'X-H})- Let 
k=yfjy^. Then h^ = /:x(xa/Xm) = CD/AC; we can determine the most productive scale size (measured 
for inputs) as x^ = {hyk)*KA. In order to determine mpss for e.g. A, we need to know h^ and k. These 
coefficients can be determined using DEA. When the DEA programme has a unique solution, we can 
apply the methodology of Banker (1984). This methodology is extended by Banker and Thrall (1992) 
to allow for allow for multiple optimal solutions. See also Appa and Yue (1999) for an analysis of 
scale efficient targets. 



■* Note that for each input-output mi.\ there is a mpss. An mpss is not necessarily the (an) optimal scale 
size; this depends on the input and output prices. 



3 Data Envelopment Analysis 

In data envelopment analysis (DEA) one uses a series of linear programming problems to draw a 
production frontier. The efficiency of each airport (or more general, a dmu) is evaluated against this 
frontier. Hence the efficiency of an airport is evaluated relative to the performance of other airports. 
More formally, assume we have L airports with m outputs and n inputs. Chames et al. (1978) propose 
the following measure of efficiency, which is the maximum of the ratio of weighted outputs to 
weighted inputs subject to the condition that for every airport the efficiency measure is smaller than or 
equal to 1 : 



max-T^ 



t.uji_c 



m 

s.t. ^ <1, l = \...L 

(1) Chames et al. (1978) show 
that the above firactional programming program has the following linear programming equivalent: 



1-1 

m n 

s.t. Iw,-:v,-,,-Zv.;:., <0, l = \...L 

1-1 >-i 

n 

«,,v,>0 

(2) 
To determine the mpss the dual to this problem is used: 

min/i(, 



s.t. Y^yu ^y-.o, i = l,.-, 



m 



T^Xji =\Xjg, j = \,..,n 
A„h,>0 



(3) 



Banker (1984) shows that a dmu represents a mpss iff /lo = 1. Moreover, by defining i^ = X'^ . the 



L 



h, 1 , 

input-output mix I -; — Xg , ~ — 3^ is a production possibility (i.e. is an element of T in figure 1) and 

is mpss. If the sum of weights /:„ > 1, local decreasing returns to scale are prevailing; if io < 1, local 
increasing returns to scale are prevailing. 

The efficiency coefficient can be either input-oriented (as in (3)) or output-oriented. If the 
input oriented coefficient /i,, < 1 in (3), it is possible to reduce all inputs (by (l-/io)xlOO%) keeping the 
outputs constant. Likewise, if the output coefficient is larger than 1 it is possible to increase the outputs 
keeping the inputs constant. 



4 Determination of efficiency coefficients 

The empirical application of this paper is undertaken in two steps. First, the DEA model is used to 
determine the efficiency coefficients and mpss. Second, the efficiency coefficients are used as 
dependent variables in a censored regression model to explain the differences in efficiency (as was 
done by Gillen and Lall (1997) for US airports). 

Data on inputs have been obtained from lATA's (1998) Airport Characteristics / Demand 
Profiles and from some airports directly. The Airport Characteristics / Demand Profiles docs not 
contain information on the number of employees. Numbers of employees can be obtained from the 
Airport Council International (ACI) (1999). Both the number of people employed by the airport 
operator and the total number of employees at the airport are available. The first number of people 
does not include people working for sub-contracting firms. For example, the number of people 
employed by the operator of FRA (Flughafen Frankfurt Main AG) in 1997 is 12,500. According to the 
ACI airport database FRA is the only airport operated by the operator. The British Airport Authority 
(BAA) had 8,393 employees and operates, amongst others, Heathrow, Gatwick and Stansted (BAA, 
1999). The numbers of passengers at these airports in 1997 were 58 million for Heathrow and 40 
million for Frankfurt. We may assume that such numerical differences reflect differences in the way 
workers have been classified in the various airports. Unfortunately, vital information is lacking on e.g. 
subcontracting, the number of people employed for aircraft handling etc. As an alternative we might 
use the total number of people employed at the airport, but then we would also include people who 
have little to do with the "airport business" and we would include too much heterogeneity^. Therefore, 
labor is not included in the analysis. This means that an efficient dmu may or may not be labor 
efficient; but by assuming zero substitutability between labor and other factors of production this 
should not be a problem. 



"' For example, before the aircraft manufacturer Fokker was closed in 1995, the emplo>Tnent of its 
Schiphol manufacturing plant was included in the airport employment figure. , 



As we only have data on runway characteristics and characteristics of passenger terminals, we 
use two output measures: air transport movements and passenger movements. Data for the period 
1995-1997 were obtained from the British Airport Authority. 

Like Gillen and Lall (1997), we analyze terminal output (PAX, the number of 
passengers) and aircraft movements (ATM, air transport movements) separately. Although the airport 
can be seen as a multi-product fum and the two outputs are clearly related, the "production 
technology" for the two outputs is quite different. Estimation results for PAX are given in tables 1 and 
2. The inputs used were terminal size (square meters), number of aircraft parking positions at the 
terminal, number of remote aircraft parking positions, number of check-in desks and number of 
baggage claims. 



Table 1 



Relative Efficiency, APM 



City 



Table 2 Sum of weights kg, APM 



Airport 



1995 



1996 



1997 



Airport 



1995 



1996 1997 RS' 



1 returns to scale characterization; 
i = increasing, d = decreasing 



Amsterdam 


AMS 


0.583 


0.639 


0.728 


AMS 


1.282 


1.407 


1.600 


d 


Berlin 


SXF 


0.372 


0.355 


0.374 


SXF 


0.106 


0.101 


0.107 


i 


Berlin 


TXL 


0.651 


0.660 


0.686 


TXL 


1.353 


1.372 


1.425 


d 


Brussels 


BRU 


0.791 


0.838 


1.000 


BRU 


0.791 


0.838 


1.000 




Bucharest 


OTP 


0.207 


0.200 


0.209 


OTP 


0.100 


0.096 


0.100 




Copenhagen 


CPH 


0.850 


0.926 


1.000 


CPH 


0.850 


0.926 


1.000 




Dublin 


DUB 


0.780 


0.881 


1.000 


DUB 


0.780 


0.881 


1.000 




Faro 


FAO 


0.997 


0.998 


1.000 


FAO 


0.997 


0.998 


1.000 




Frankfurt 


FRA 


0.815 


0.828 


0.857 


FRA 


1.694 


1.720 


1.780 


d 


GSteborg 


GOT 


0.442 


0.469 


0.496 


GOT 


0.201 


0.213 


0.225 




Hamburg 


HAM 


0.484 


0.485 


0.510 


HAM 


0.463 


0.464 


0.488 




Hannover 


HAJ 


0.473 


0.493 


0.535 


HAJ 


0.260 


0.271 


0.294 




Lamaca 


LCA 


0.647 


0.655 


0.659 


LCA 


0.155 


0.156 


0.157 




Leeds/Bradford LBA 


0.466 


0.366 


0.461 


LBA 


0.063 


0.073 


0.084 




Lisbon 


LIS 


0.795 


0.844 


0.877 


LIS 


0.966 


1.026 


1.066 


i/d 


London 


LGW 


0.785 


0.845 


0.940 


. LGW 


0.613 


0.660 


0.734 




London 


LHR 


0.936 


0.964 


1.000 


LHR 


0.936 


0.964 


1.000 




London 


STN 


0.347 


0.430 


0.479 


STN 


0.220 


0.272 


0.304 




Lyon 


LYS 


0.367 


0.411 


0.398 


LYS 


0.254 


0.283 


0.275 




Manchester 


MAN 


0.645 


0.634 


0.687 


MAN 


0.947 


0.930 


1.009 


i/d 


Marseille 


MRS 


0.873 


0.956 


0.987 


MRS 


0.314 


0.332 


0.342 




Milan 


LIN 


0.744 


0.869 


1.000 


LIN 


0.744 


0.869 


1.000 




Milan 


MXP 


0.315 


0.311 


0.325 


MXP 


0.234 


0.231 


0.241 




Munich 


MUC 


0.570 


0.602 


0.687 


MUC 


0.982 


1.037 


1.184 


i/d 


Nuremberg 


NUE 


0.424 


0.419 


0.455 


NUE 


0.066 


0.065 


0.071 


i 


Paris 


CDG 


0.466 


0.523 


0.584 


CDG 


1.276 


1.433 


1.600 


d 


Paris 


ORY 


0.798 


0.821 


0.752 


ORY 


1.68S 


1.736 


1.589 


d 


Prague 


PRG 


0.290 


0.339 


0.373 


PRG 


0.189 


0.221 


0.243 


i 


Rome 


FCO 


0.782 


0.854 


0.927 


FCO 


1.193 


1.303 


1.414 


d 


Stockhohn 


STO 


0.777 


0.805 


0.866 


STO 


0.801 


0.830 


0.893 




Smttgart 


STU 


0.695 


0.870 


0.922 


STU 


0.352 


0.440 


0.467 




Turin 


TRN 


0.195 


0.217 


0.259 


TRN 


0.104 


0.116 


0.138 




Vienna 


VIE 


0.864 


0.924 


0.990 


VIE 


0.496 


0.531 


0.569 




Zurich 


ZRH 


0.835 


0.884 


1.000 


ZRH 


0.835 


0.884 


1.000 





From table 1 it appears that the relative efficiency measure for most airports increases 
(slightly) over time. The regional dispersion remains more or less constant over time. There are large 
differences in efficiency between airports in given years, and even among cities. For example, the 
efficiency coefficient of London Stansted (STN) is much lower than the coefficient for London 
Heathrow (LHR). The same observation holds true for Flughafen Berlin-Schonefeld (SXF) and 
Flughafen Berlin-Tegel (TXL). From table 2 it appears that most airports are operating under local 
increasing returns to scale. Some airports (BRU, CPH, DUB, FAO, LHR, LIN and ZRH) have been 
increasing there scale of operation such that they are efficient (and mpss) in 1997 compared to 
previous years and other to airports. Other airports (GOT, HAM, HAJ, LCA, LBA, LGW, STN, LYS, 
MRS, MUC, MXP, NUE, OTP, PRO, STO, SXF, TRN, VIE) have been increasing their efficiency 
over time, but are still inefficient (are not yet mpss) compared to other airports. These airports could 
increase their scale to reach mpss. AMS, TXL, FRA, CDG and DRY are operating under local 
decreasing returns to scale. Given their current input combinations, these airports could decrease their 
scale to reach mpss. As was explained in section 2, this mpss may not be the optimal scale of 
operations for an airport; this depends on the prevailing output and input prices. In fact, changing the 
input mix also changes the mpss, and with it, the sign of the returns to scale may change for the years 
in which these airports did not yet reach mpss: with a more favorable input mix AMS, TXL, FRA, 
CDG and ORY could improve their position. As was mentioned in the introduction, DBA only 
measures inefficiency and does not explain. Clearly, more research is needed to explain the 
unfavorable positions of AMS, TXL, FRA, CDG and ORY and to investigate whether expansion of 
these airports is economically justified. LIS, MAN and MUC are operating near the mpss (/:„ is just 
below 1 or just above 1). The relative inefficiency coefficients appear to be on the low side for airports 
so operating so close to the mpss. This suggests that at these three airports technical inefficiency 
dominates scale inefficiency. 

The inefficiency ratios for ATM are given in table 3. The inputs used are the total 
airport area (ha), total length of the runway system, number of aircraft parking positions at the terminal 
and the number of remote parking positions. It appears that, again, over time relative efficiency 
increases. It appears there are fewer airports achieving relative efficiency (and mpss) in ATM then 
there are airports reaching relative efficiency (and mpss) in PAX. The observations on the Berlin and 
London airports made in table 1 also appear in table 3. From table 4, it appears that most airports are, 
again, operating under local increasing returns to scale, but apart from CDG, CPH, LGW, LHR, LIN 
and STU no airport reaches mpss. AMS, BRU, FCO, FRA, MUC, ORY, STO and ZRH are operating 
under local decreasing retums to scale. 

The conclusion so far is that for both outputs, most airports are operating under increasing 
retums to scale. This indicates that, to improve relative efficiency, most airports could increase their 
scale of operations to reach mpss, or already have done so. Some airports (AMS, FCO, FRA and ORY) 
could decrease their scale to achieve the mpss at the current input mix. Finally, BRU, MUC and ZRH 
operate under local increasing retums to scale when looking at PAX, while they are operating under 
decreasing returns to scale looking at AT.M. The opposite holds true for CDG and TXL. 



Table 3 



Relative Efficiency, ATM 



City 



Airport 



1995 



1996 



Table 4 Sum of weigfits kn, A TM 



1997 



Airport 



1995 



1996 



1997 RS' 



Amsterdam 


AMS 


0.616 


0.682 


0.741 


AMS 


1.408 


1.558 


1.692 d 


Berlin 


SXF 


0.155 


0.160 


0.144 


SXF 


0.165 


0.170 


0.153 i 


Berlin 


TXL 


0.469 


0.488 


0.490 


TXL 


0.784 


0.816 


0.819 i 


Brussels 


BRU 


0.607 


0.661 


0.697 


BRU 


1.309 


1.425 


1.503 d 


Bucharest 


OTP 


0.154 


0.132 


0.146 


OTP 


0.191 


0.164 


0.182 i 


Copenhagen 


CPH 


0.837 


0.937 


1.000 


CPH 


0.837 


0.937 


1.000 i 


Dublin 


DUB 


0.355 


0.353 


0.389 


DUB 


0.499 


0.497 


0.547 i 


Faro 


FAO 


0.306 


0.290 


0.297 


FAO 


0.142 


0.134 


0.138 i 


Frankfurt 


FRA 


0,905 


0.921 


0.932 


FRA 


2.134 


2.170 


2.197 d 


Goteborg 


GOT 


0.545 


0.548 


0.564 


GOT 


0.345 


0.346 


0.357 d 


Hamburg 


HAM 


0.535 


0.542 


0.562 


HAM 


0.502 


0.508 


0.527 i 


Harmover 


HAJ 


0.438 


0.438 


0.564 


HAJ 


0.301 


0.301 


0.388 i 


Lamaca 


LCA 


0.262 


0.259 


0.266 


LCA 


0.192 


0.189 


0.194 i 


Leeds/Bradford LBA 


0.567 


0.549 


0.580 


LBA 


0.189 


0.183 


0.193 i 


Lisbon 


LIS 


0.424 


0.484 


0.465 


LIS 


0.408 


0.466 


0.448 i 


London 


LGW 


0.837 


0.920 


1.000 


LGW 


0.837 


0.920 


1.000 i 


London 


LHR 


0.976 


0.995 


1.000 


LHR 


0.976 


0.995 


1.000 i 


London 


STN 


0.410 


0.481 


0.523 


STN 


0.304 


0.356 


0.388 i 


Lyon 


LYS 


0.289 


0.327 


0.361 


LYS 


0.457 


0.516 


0.570 i 


Manchester 


MAN 


0.749 


0.718 


0.746 


MAN 


0.540 


0.518 


0.539 i 


Marseille 


MRS 


0.427 


0.468 


0.491 


MRS 


0.258 


0.283 


0.297 i 


Milan 


LIN 


0.802 


0.949 


1.000 


LIN 


0.802 


0.949 


1.000 i 


Milan 


MXP 


0.319 


0.285 


0.302 


MXP 


0.242 


0.216 


0.229 i 


Munich 


MUC 


0.754 


0.826 


0.957 


MUC 


1.228 


1.344 


1.557 d 


Nuremberg 


NUE 


0.473 


0.476 


0.506 


NUE 


0.159 


0.159 


0.170 i 


Paris 


CDG 


0.823 


0.912 


1.000 


CDG 


0.823 


0.912 


1.000 i 


Paris 


ORY 


0.654 


0.690 


0.666 


ORY 


1.297 


1.367 


1.321 d 


Prague 


PRG 


0.323 


0.348 


0.364 


PRG 


0.274 


0.295 


0.309 i 


Rome 


FCO 


0.697 


0.788 


0^819 


FCO 


1.151 


1.301 


1.352 d 


Stockhohn 


STO 


0.791 


0.835 


0.904 


STO 


1.163 


1.227 


1.329 d 


Stuttgart 


STU 


0.727 


0.810 


1.000 


STU 


0.727 


0.810 


1.000 i 


Turin 


TRN 


0.216 


0.239 


0.280 


TRN 


0.152 


0.169 


0.198 i 


Vieima 


VIE 


0.570 


0.614 


0.618 


VIE 


0.884 


0.952 


0.958 i 


Zurich 


ZRH 


0.851 


0.914 


0.983 


ZRH 


1.015 


1.090 


1.173 d 



1 returns to scale characterization; 
i = increasing, d = decreasing 

The question arises whether the conjecture that returns to scale are relatively stronger at 
relatively smaller airports is supported by the analysis. There is one airport (TXL) operating under 
local decreasing returns to scale which is relatively small, and there are 5 relatively large airports 
(FRA, CDG, ORY, AMS, FCO) that also operate under decreasing returns to scale that do not fit this 
pattern. Given the different patterns form these 6 airports, we feel the correlation coefficient behveen 
APM and ^^ reported in table 2 is sufficiently high (0.718) to provide some support for the conjecture 
that low values of l<o (high returns to scale) comcide with relatively high values of .A.PM. The same 



holds true for ATM, where the correlation coefficient (between ATM and kg reported in table 4) is even 
higher (0.863). 

Using the efficiency coefficients and sums of weights reported in tables 1-4, we can determine 
the mpss for these airports as indicated in sections 2 and 3. As already mentioned, these are the mpss 
given the current input proportions; it might be better for an airport to adjust the input ratio rather than 
to change all inputs proportionally. Without data on input prices, however, it is difficult to say anything 
about the "optimal mpss". Moreover, given the indivisibilities, interpreting the mpss may be difficult. 
This becomes clear in table 6, appendix 2, where, for example, LBA should construct no less than 6 
runways with an average length of 1675 meters length to reach the mpss. With an average length of 
2930 meters the number reduces to 3, which is more in line with the APM at the mpss. 

From table 5 it appears that a number of airports (GOT, HAJ, HAM, LBA, LCA, LGW, LYS, 
MRS, MXP, NUE, OTP, PRO, STN, STU, SXF, TRN and VIE) have a large growth potential 
compared to the other airports at the current input mix. AMS, CDG, ORY and TXL on the other hand 
have no growth potential at the current input mix. 

Note that while TXL could decrease its size, the other Berlin airport, SXF, could increase its scale. A 
similar observation is made for the London airports, where LGW and STN could increase their scale of 
operations and LHR should remain constant. 

Changing the input mix changes the results; for example reducing the total length of the 
runway system and the number of parking positions at AMS (so that they are proportional to LHR) but 
keeping the airport area constant renders AMS efficient in 1997 and changes the sign of the returns to 
scale'. From an economic perspective, this may be a better position for AMS than the mpss reported in 
table 6. The mpss reported in tables 5 and 6 are technically efficient, but not necessarily cost efficient. 
To determine the "true" optimal scale of operations, also cost data are needed. These are not have 
available. 

A final observation made on the basis of tables 5 and 6 is that the number of aircraft parking 
positions at the PAX mpss is, on average, larger than the number of aircraft parking positions at the 
ATM mpss; a larger number of parking positions is needed to handle passengers than is needed to 
accommodate arriving and departing aircraft. Second, more inputs may be needed. For example, in 
this model, the airport which has e.g. the smallest terminal size compared to the number of passengers 
is (can be) the most efficient. However, a larger terminal allows for more amenities at the aiiport, 
which may be needed to attract passengers. 



5 Stochastic Frontier Analysis 

In this section, the results of the previous section will be compared to the results of a stochastic frontier 
analysis. Consider the following stochastic production frontier: 



' AMS area; 2200, runway length: 524S, terminal positions: 72, remote positions: 37. 



10 



yj., = -^y., ^+ ^j, 

(5) 
where y^, is the output of airporty in period / and x^, are the inputs of aiiporty in period t. Both;' and x 
can be defined in terms of the original units of productions or in logarithms. UpH{Q,uJ-) and IID and is 
also independent of Vj„ which is distributed according to half normal distribution with variance 0-'/. U 
is the (stochastic) deviation from the production frontier; for Uj > airport j does not reach the 
(efficient) frontier. The technical efficiency of airport; is' (see also Battese and Coelli (1988)): 



E 



(6) 
where the superscript denotes that the efficiency coefficient is obtained from a frontier model. If 
equation (5) is specified in logs, them A/= exp(f^). 

Note that the stochastic frontier model in equation (5) can be extended to explam the 
inefficiencies Uj. The firontier model and the inefficiency model Uj = UJ^Zj) are estimated 
simultaneously. Unfortunately, we do not have the necessary variables Zj to explain the inefficiencies. 

Both the models explaining ATM and APM were specified in logs. Not all variables used in 
the DEA could be used to estimate a stochastic production frontier'. For the model explaining APM, 
The explanatory variables included are, next to a constant, the number of baggage claim units claims, 
the number of parking positions at the terminal {pos) and the number of remote parking positions pos. 
The estimates are (standard error between parentheses): 

OLS: 

]n{APM)= 5.907 + Q.Z26hx{claims) + 02(t\-\n[pos) + 0.165-hi(reffi) 

(0.297) (0.097) (0.060) (0.105) 

ML: 

h(^PyVO= 6.471 + 0.771-bi(c/aim5) + 0.253-hi(poi) + 0.201-hi(rem) 

(0.384) (0.099) (0.058) (0.110) 

Log{likelihood) = -76.47 /= o-„Vo;' = 2.17 (1.15) 

The null hypothesis ;k = is rejected at the 90% confidence level. Note that, as y is larger than 2, the 
variance of Uj is large compared to the variance of Vj„ indicating there is a substantial inefficiency 
effect. The variables explaining ATM are: ninways is the number of runways, pos is the number of 
aircraft parking positions at the terminal and rem is the number of remote parking positions. 



' There are other distributions for Kpossible, see e.g. Battese and Coelli (1988). 
' Note that, again, a technically efficient airport may operate cost inefficiently. 



OLS: 

ATM= 1.777 + 0.50Mn(n/nvra>'j) + 0.3431n(poj) + 0.435-ln(re;n) 

(0.311) (0.140) (0.059) (0.103) 

ML: 
.4 7^= 2.426 + 0.4041n(rx/«wq>'j) + 0.3l4\n{pos) + 0.472 •In(reffi) 

(0.479) (0.134) (0.048) (0.115) 

Log(JikeUhood) = -82. 10 /= olVo;' = 2.60 (0.08) 

Again, the null hypothesis ;' = is rejected at the 90% confidence level and there is a substantial 
inefficiency effect. 

It is noted that a yearly dununy D may be necessary (as we have pooled time series - cross 
section data), but taking this into account in the estimations does not change the results. Moreover, the 
estimation results arc not really robust. Including additional variables may necessitate a different 
specification of the production function. Also, the null hypothesis / = is rejected at the 90% 
confidence level, but is not rejected at 95% confidence. 

The efficiency coefficients hj are reported in appendix 3. Despite the fact that there is 
somewhat less temporal and regional dispersion of the efficiency coefficients using stochastic frontier 
models than they are using DEA (the error term in equation (5) consists of the "inefficiency" Uj and 
the "noise" Vj„ which is "filtered out"'), the stochastic efficiency frontier model seems to be able to 
reproduce the DEA results quite reasonably. Notable exceptions are OTP and MXP which perform 
much better when looking at the stochastic frontier model, and TXL, MXP and ORY which perform 
much worse under the stochastic frontier model. 

Although the estimations do not seem to be very robust, the stochastic frontier model quite 
reasonably reproduces the DEA results. Future research, using more flexible specifications, should 
verify this result. 



Conclusion 

In this paper efficiency indices and most productive scale sizes were determined for European airports. 
Most airports seem to be operating under increasing returns to scale. Large differences in relative 
efficiency exist, and also large deviations from the mpss are found. It should be kept in mind that the 
mpss are determined given the current input-mix. A change in the input mix may lead to different 
outcomes. 



' A less restrictive assumption on the distribution of t/y would be that Uf\^{ji,<j^), truncated at 0. 
Including all variables, the null hypotheses //= and a„- are, however, not rejected; Uj cannot be 
distinguished from 0. 

' It is noted that care should be taken comparing the results of a non-parametric and a parametric 
approach. ; 



12 



Some airports, like AMS, are operating under local decreasing returns to scale. As a result, the 
mpss lies well below the current scale of operations. This indicates that a reduction in the scale of 
operations (both in inputs and outputs) would be wise, but a change in the input mix could be more 
advisable. For example, using the input mix of LHR, but then proportionally to the demand at AMS 
(see footnote 8), results in a mpss with a higher out put then the mpss reported in tables 6 (actually, the 
mpss for ATM is the same as realized output for 1997). Thus, from this analysis of production 
efficiency, we can conclude that i) AMS has probably not an optimal input mix (even at the mpss 
reported in table 6) and ii) in any case, the planned construction of a fifth runway is not a good idea, 
from a pure economic point of view. But, as already mentioned in the introduction, this analysis was 
primarily focused on the measurement on inefficiency. In the case of AMS, the inefficiency is, for a 
large part, caused by regulations. This becomes clear from figure 2. The politically instigated noise 
contours are "chosen" such that the overlap with population centers is minimized. The expected growth 
in the number of air transport movements (ATM) will increase the noise (and safety) problem. Given 
the regulation, the expected growth apparently cannot be accommodated at the existing runway system. 
Hence a fifth runway is planned. Form a pure economic point of view, this is, as akeady said, not the 
optimal solution. 

The stochastic frontier analysis seems to produce reasonable efficiency coefficients and might 
be considered more flexible than DEA as it includes a "noise" term; the inefficiency is the distance to 
the frontier p/«5 a random error term rather than to the (different) frontier itself as in DEA. The use of a 
more flexible fiinctional form of the production fiinction may however be necessary if this method is 
used to measure and explain inefficiency. 

The research agenda that follows from this paper is the following. First and foremost, more 
attention has to be paid to the "explaining" of inefficiency, either using a stochastic frontier model or 
DEA output. Second, as the mpss is not necessarily the optimal scale of an airport, an (empirical) 
analysis of airport cost functions should shed more light on the "true" optimal scale of an airport; this 
is the mpss at the cost minimizing input mix. 



Acknowledgments 

The authors wish to thank the British Airport Authority for making available the data. 

References 

Airport Council International (ACI) (1999), Airport Database. 

Appa, G. and M. Yue (1999), On setting scale efficient targets in DEA, Journal of the Operational 

Research Society, 50, 60-69. 
British Airport Authority (1999), Annual report 1997/98. 
Banker, R.D. (19S4), Estimating most productive scale size using data en\elopment analysis, 

European Journal of Operational Research, 17, 35-44. 



13 



Banker, R.D. and R.M. Thrall (1992), Estimation of returns to scale using data envelopment analysis, 

European Journal of Operational Research, 62, 74-84. 
Button, K.J. and T.G. Weyman- Jones (1994), X-efficiency and technical efficiency, Public Choice, 80, 

83-104. 
Chames, A., W.W. Cooper and E. Rhodes (1978), Measuring the efficiency of decision making units, 

European Journal of Operational Research, 2, 429-444. 
Gillen, D. and A. Lall (1998), Non-Parametric Measures of Efficiency of U.S. Airports, paper 

presented at the 2""* Air Transport Research Group Symposium, Dublin, 20-21 July 1998. 
Gillen, D. and A. Lall (1997), Developing Measures of Airport Productivity and Performance: an 

Application of Data Envelopment Analysis, Transportation Research, 33E(4), 261-274. 
Hooper, P.G. and D.A. Henscher (1997), Measuring Total Factor Productivity of Airports: an Index 

Number Approach, Transportation Research, 33E(4), 249-260. 
International Air Transport Association (1998), Airport Capacity / Demand Profiles. 
Oum, T.H., W.G. Waters II and C. Yu (1998), A survey of productivity and efficiency measurement in 

rail transport, Journal of Transport Economics and Policy, 33(1), 9-42. 
Tolofari, S., N.J. Ashford and R.E. Caves (1990), The Cost of Air Service Fragmentation, Report 

TT9010, Department of Transport Technology, Loughborough University of Technology. 



14 



Appendix 1 Airports used in the Analysis 

AMS Amsterdam Airport Schiphol LIS 

BRU Brussels National LYS 

CDG Charles de Gaulle MRS 

CPH Copenhagen International Airport MUC 

DUB Dublin NUE 

FAO Faro ORY 

FCO Leonardo da Vinci OTP 

ERA Frankfurt Main International PRG 

GOT Goteborg-Landvetter Airport STN 

HAJ Hannover STU 

HAM Hamburg International SXF 

LEA Leeds-Bradford International Airport TRN 

LCA Lamaca TXL 

LGW London Gatwick VIE 

LHR London Heathrow ZRH 



Lisbon International 

Aeroport de Lyon-Satolas 

Aeroport International Marseille-Provence 

Flughafen Munchen 

Flughafen Nuremberg 

Orly 

Otopeni International 

Ruzyne 

London Stansted 

Flughafen Stuttgart 

Flughafen Berlin-Schonefeld 

Citta' Di Torino 

Flughafen Berlin-Tegel 

Vienna International 

Zurich International 



Appendix 2 MPSS and Data, 1997 



Table 5 


MPSS, PAX 
















terminal aircraft 


parking positions chcck- 


in baggage 






PAX 


size (m') positions remote 


desks 


claims 




AMS 


19382 


168197 


30 


27 


151 


9 


BRU 


15935 


125000 


32 


66 


143 


5 


CDG 


21933 


161292 


23 


34 


157 


9 


CPH 


16837 


97000 


49 


14 


84 


7 


DUB 


10235 


37889 


50 


37 


99 


5 


FAO 


3664 


4400 


6 


13 


31 


5 


FCO 


17684 


114753 


16 


31 


119 


8 


FRA 


22542 


338776 


24 


35 


194 


16 


GOT 


15097 


97030 


18 


42 


90 


7 


HAJ 


15922 


72818 


22 


33 


95 


22 


HAM 


17530 


69530 


67 


37 


101 


10 


LBA 


14786 


208500 


7 


119 


194 


22 


LCA 


23345 


87976 


121 


121 


168 


13 


LOW 


36508 


184806 


67 


38 


287 


20 


LHR 


57808 


211400 


104 


53 


482 


30 


LIN 


14395 


67400 


5 


31 


93 


6 


LIS 


6223 


19423 


6 


24 


77 


7 


LYS 


17530 


69541 


33 


72 


101 


14 


MAN 


15574 


49932 


40 


22 


140 


11 


MRS 


16280 


235757 


81 


45 


178 


6 


MUC 


14891 


109755 


14 


30 


91 


7 


MXP 


14609 


70971 


7 


31 


93 


7 


NUE 


33981 


119050 


90 


38 


269 


19 


OTP 


14674 


83235 


10 


85 


92 


8 


ORY 


15748 


175757 


17 


27 


98 


8 


PRO 


16755 


115726 


48 


25 


84 


9 


STN 


17678 


77348 


47 


33 


133 


8 


STO 


16742 


123382 


97 


124 


90 


7 


STU 


14446 


84750 


6 


69 


93 


12 


SXF 


17537 


69234 


■ 81 


109 


102 


11 


TRN 


17232 


81343 


58 


49 


94 


9 


TXL 


6049 


13062 


25 


41 


44 


8 


VIE 


16867 


55694 


35 


85 


146 


10 


ZRH 


17871 


56000 


24 


30 


110 


12 



16 



Table 6 MPSS, ATM 







Aiiport 




Runway aircraft 


parking positions 




ATM 


Area (ha) Runway; 


i 


Length (m) termina 


1 remote 




AMS 


207 


963 


2 


5890 


28 


26 


BRU 


169 


577 


1 


4559 


15 


31 


CDG 


396 


3109 


2 


7800 


63 


93 


CPH 


284 


1240 


3 


8665 


49 


14 


DUB 


222 


768 


2 


4275 


36 


26 


FAO 


186 


1084 


2 


5367 


13 


28 


FCO 


182 


969 


2 


6722 


15 


28 


FRA 


177 


806 


1 


5090 


21 


31 


GOT 


170 


1186 


2 


5219 


13 


30 


HAJ 


199 


896 


4 


10067 


17 


26 


HAM 


241 


601 


2 


7369 


68 


37 


LBA 


135 


429 


6 


10050 


3 


48 


LCA 


182 


397 


1 


3698 


40 


40 


LGW 


229 


759 


1 


3098 


52 


30 


LHR 


429 


1200 


2 


7560 


104 


53 


LIN 


165 


360 


2 


2440 


5 


31 


LIS 


171 


522 


2 


6445 


7 


30 


LYS 


165 


696 


1 


4222 


15 


32 


MAN 


276 


831 


1 


4224 


82 


44 


MRS 


280 


991 


3 


9744 


48 


26 


MUC 


164 


921 


1 


4914 


15 


32 


MXP 


170 


1611 


3 


9946 


7 


30 


NUE 


264 


1075 


3 


8059 


42 


18 


OTP 


180 


778 


2 


4728 


18 


29 


ORY 


150 


617 


2 


5627 


4 


33 


PRG 


250 


1062 


4 


10694 


36 


19 


STN 


218 


1293 


1 


4113 


40 


28 


STO 


194 


435 


1 


3946 


68 


87 


STU 


135 


240 


1 


3345 


3 


35 


SXF 


200 


598 


2 


5363 


22 


29 


TRN 


190 


425 


1 


4680 


44 


37 


TXL 


145 


278 


1 


3256 


30 


51 


VIE 


163 


645 


1 


4582 


13 


32 


2RH 


206 


677 


3 


7966 


20 


25 



17 



Table? Data, 1997 







Airport 




Runway Terminal Aircraft parking positions Check-in Baeeaee 




PAX ATM Area Runways 1 


Length Size terminal remote 


desks claims 




AMS 


31021 


349.5 


2200 


4 


13450 


370000 


65 


60 


333 


19 


BRU 


15935 


254.7 


1245 


3 


9833 


125000 


32 


66 


143 


5 


CDG 


35103 


395.5 


3109 


2 


7800 


442200 


63 


93 


430 


26 


CPH 


16837 


283.6 


1240 


3 


8665 


97000 


49 


14 


84 


7 


DUB 


10235 


121.3 


1080 


3 


6010 


37889 


50 


37 


99 


5 


FAO 


3664 


25.6 


503 


1 


2490 


4400 


6 


13 


31 


5 


FCO 


25001 


245.7 


1600 


3 


11095 


175000 


25 


47 


182 


12 


FRA 


40128 


389.6 


1900 


3 


12000 


704000 


49 


73 


403 


34 


GOT 


3394 


60.8 


750 


1 


3300 


44000 


8 


19 


41 


3 


HAJ 


4676 


77.1 


616 


3 


6920 


40000 


12 


18 


52 


12 


HAM 


8546 


127.1 


564 


2 


6910 


66500 


64 


35 


97 


10 


LBA 


1247.7 


26.1 


143 


2 


3350 


28000 


1 


16 


26 


3 


LCA 


3673 


35.3 


290 


1 


2700 


21000 


29 


29 


40 


3 


LOW 


26795 


229.3 


759 


1 


3098 


144361 


52 


30 


224 


16 


LHR 


57808 


429.2 


1200 


2 


7560 


211400 


104 


53 


482 


30 


LIN 


14395 


165.3 


360 


2 


2440 


67400 


5 


31 


93 


6 


LIS 


6631 


76.8 


503 


2 


6205 


23598 


7 


29 


93 


8 


LYS 


4819 


94.1 


1100 


2 


6670 


48000 


23 


50 


70 


10 


MAN 


15714 


148.5 


600 


1 


3048 


73300 


59 


32 


206 


16 


MRS 


5574 


83.3 


600 


2 


5900 


84750 


29 


16 


64 


2 


MUC 


17626 


256 


1500 


2 


8000 


189000 


24 


52 


157 


12 


MXP 


3523 


38.8 


1220 


2 


7530 


52700 


5 


23 


69 


5 


NUE 


2418 


44.8 


360 


1 


2700 


18600 


14 


6 


42 


3 


OTP 


1470.7 


27.3 


768 


2 


7000 


40000 


5 


41 


44 


4 


DRY 


25023 


237.1 


1541 


3 


9370 


371500 


36 


58 


207 


16 


PRO 


4078 


77.3 


902 


3 


9085 


75500 


31 


16 


55 


6 


STN 


5366.6 


84.4 


958 


1 


3048 


49000 


30 


21 


84 


5 


STO 


14951 


257.4 


640 


2 


5800 


127205 


100 


128 


93 


7 


STU 


6745 


134.9 


240 


1 


3345 


42918 


3 


35 


47 


6 


SXF 


1870 


30.7 


636 


2 


5700 


19760 


23 


31 


29 


3 


TRN 


2377 


37.6 


300 


1 


3300 


43300 


31 


26 


50 


5 


TXL 


8622 


119.1 


465 


2 


5447 


27150 


51 


86 


92 


17 


VIE 


9597 


155.9 


1000 


2 


7100 


32000 


20 


49 


84 


6 


ZRH 


17871 


241.5 


807 


3 


9500 


56000 


24 


30 


110 


12 



IS 



Appendix 3 Frontier efficiency measures 

Table 8, efficiency coefficients from frontier analysis 

ATM 



APM 



95 



96 



97 



95 



96 



97 



Amsterdam 


AMS 


0.600 


0.643 


0.677 


0.633 


0.668 


0.714 


Berlin 


SXF 


0.206 


0.211 


0.193 


0.350 


0.337 


0.351 


Berlin 


TXL 


0.320 


0.332 


0.333 


0.281 


0.285 


0.294 


Brussels 


BRU 


0.610 


0.646 


0.668 


0.797 


0.810 


0.843 


Bucharest 


OTP 


0.641 


0.663 


0.648 


0.756 


0.764 


0.738 


Copenhagen 


CPH 


0.830 


0.851 


0.861 


0.813 


0.830 


0.843 


Dublin 


DUB 


0.384 


0.383 


0.416 


0.655 


0.700 


0.741 


Faro 


FAO 


0.446 


0.425 


0.434 


0.641 


0.642 


0.643 


Frankfurt 


FRA 


0.747 


0.753 


0.757 


0.638 


0.643 


0.657 


Goteborg 


GOT 


0.668 


0.670 


0.682 


0.663 


0.685 


0.705 


Hamburg 


HAM 


0.458 


0.462 


0.477 


0.433 


0.433 


0.451 


Hannover 


HAJ 


0.447 


0.447 


0.551 


0.351 


0.363 


0.388 


Lamaca 


LCA 


0.265 


0.263 


0.269 


0.570 


0.575 


0.577 


Leeds/Bradford LBA 


0.506 


0.493 


0.516 


0.421 


0.467 


0.533 


Lisbon 


LIS 


0.555 


0.613 


0.596 


0.616 


0.640 


0.654 


London 


LOW 


0.796 


0.819 


0.836 


0.715 


0.739 


0.770 


London 


LHR 


0.799 


0.804 


0.805 


0.749 


0.758 


0.768 


London 


STN 


0.521 


0.589 


0.626 


0.467 


0.550 


0.594 


Lyon 


LYS 


0.348 


0.387 


0.421 


0.308 


0.337 


0.329 


Manchester 


MAN 


0.691 


0.674 


0.689 


0.537 


0.529 


0.562 


Marseille 


MRS 


0.500 


0.539 


0.559 


0.821 


0.832 


0.838 


Milan 


LIN 


0.813 


0.847 


0.856 


0.856 


0.876 


0.891 


Milan 


MXP 


0.426 


0.386 


0.407 


0.587 


0.582 


0.600 


Munich 


MUC 


0.720 


0.751 


0.794 


0.673 


0.693 


0.737 


Niiremberg 


NUE 


0.680 


0.682 


0.706 


0.582 


0.577 


0.611 


Paris 


CDG 


0.682 


0.721 


0.752 


0.551 


0.598 


0.642 


Paris 


ORY 


0.247 


0.217 


0.236 


0.313 


0.303 


0.315 


Prague 


PRG 


0.408 


0.434 


0.452 


0.366 


0.415 


0.449 


Rome 


ECO 


0.684 


0.730 


0.743 


0.790 


0.810 


0.827 


Stockhobn 


STO 


0.409 


0.429 


0.459 


0.589 


0.604 


0.633 


Stuttgart 


STU 


0.831 


0.851 


.0.881 


0.703 


0.772 


0.787 


Turin 


TRN 


0.232 


0.254 


0.292 


0.236 


0.259 


0.299 


Vienna 


VIE 


0.618 


0.649 


0.652 


0.686 


0.709 


0.732 


Zurich 


ZRH 


0.763 


0.784 


0.803 


0.719 


0.737 


0.773 



19 



Figure 2 



Noise contours at Schiphol' " 



proposed 
fiflh runway 




40 Ke 
35 Ke 



40 Ke 
35 Ke 



1) source; PMMS (1996) 

2) Ke means "Kosten eenheid", a function of the luuiiber of decibels. As a rule, (Ke-10)% of the 
population li\ing in an area is seriously affected by airciafi noise. 



A Critical Examination of an Airport Noise 

Mitigation Scheme and an Aircraft Noise Charge: 

The Case of Capacity Expansion and Externahties at 

Sydney (Kingsford Smith) Airport 

Giovanni Nero and John A. Black * 

School of Civil and Environmental Engineering, 

University of New South Wales, 

Sydney, 2052, Australia. 

May 30, 1999 



Abstract 

In the wake of the Australian airline liberalization in 1990 and its forecasted 
impact on air traffic, capacity has been expanded at Sydney (Kingsford Smith) 
Airport [Sydney KSA] - Australia's busiest commercial airport - with the con- 
struction of the third runway in 1994. Coinciding with the approval for this ca- 
pacity expansion, the Commonwealth Government amended the Federal Airports 
Corporation (FAC) Act to direct the FAC to carry out activities which protect the 
environment from the effects of aircraft operations, with the cost to be borne by 
the airline industry according to the 'Polluter Fays Principle'. Noise management 
plans were part of the conditions for developmental approval for a third runway. 
To this end, since 1995, Sydney KSA imposes a noise levy designed to gener- 
ate sufficient revenues to fund a noise mitigation scheme. Although the issues of 
aircraft noise, in particular its impact on property values and land use planning 
around the airport, have been extensively addressed in the literature, no one has 
empirically examined the implications of new environmental policies in conjunc- 
tion with airline liberalisation and change in airport infrastructure. Principles and 
policy analyses are discussed in this paper. By focusing on the specifics of Sydney 
KSA, broader policy issues likely to be relevant for other major airports around 
the world are discussed. 

Key words: .A.irline Liberalisation, .Airport Capacity, Environmental Impact, 
Aircraft Noise, Noise Levy. 



"gio.nero@workmail.com and j.black'.Sunsw. edu.au, respectively. PREL1.VII.N'.\RY DRAFT. All com- 
ments welcome. Dr Nero thauL's the Swiss .N'ational Science Foundation for financial support (Gran: 
no. S-2 10-50417). 



in Section 5, in which we also examine the properties of this charge against OECD (1991) 
criteria and set out potential limitations of the N'LC formula used at Svdnev KSA 

The impacts of the NLC on aircraft operations and on its customers are examined in 
Section 6. Specifically, we take a range of demand elasticities and estimate the effects of 
differing levels of NLC. Aircraft noise has profound social impacts, especially on those 
living near the airport, and runway usage and flight paths determine the spatial distribu- 
tion of noise. Section 7 examines these distribution consequences - both for the parallel 
runway operations and for the current policy of 'sharing aircraft noise' encapsulated in 
AirServices Australia Long Term Operating Plan (LTOP) of 1996. Finally, in Section 8 
our conclusions are set out. 



2 Deregulation and Traffic Growth at Sydney KSA 

A growth in air traffic volume can be induced by a combination of several forces: demand 
factors (e.g. increase of GDP), supply factors (e.g. a change in the industry structure 
following a merger), and institutional factors (e.g. policy changes bringing deregulation 
or liberalization). Environmental constraints are likely to arise at any airport experienc- 
ing growth in traffic volume. In a related paper Nero and Black (199S) have argued that 
the problem of environmental externalities is exacerbated by hub development and that. 
to some extent, hubbing contributes to a spatial redistribution of externalities. Since 
the Australian airline industry has experienced major changes in competition policy this 
last decade, a somewhat detailed structural analysis of the impact of deregulation on 
Sydney SK.A is needed at this points To this end, Table 1 presents a brief summary of 
the major historical events that have shaped the Australian airline industry, and that 
have influenced, to some extent, Sydney KS.A development. In terms of competition 
policy the major event is the deregulation of the domestic market in November 1990. 



Insert Table 1 here. 



Within the context of the above events, the following analysis shows the extent to 
which Sydney KS.A has retained its role and importance as the primary .Australian 
domestic and international gateway. Table 2 and Table 3 show that Sydney KS.A is bv 
far the largest airport in Australia. Throughout the 90s Sydney KS.A passengers marke', 
share has been fairly stable, although its siiare of international aircraft movements hai 
been recentlv eroded bv Brisbane .-Virnorr,. 



Inser'. Table 2 aiici Table o iiere. 

"For a pafiial .aisessmreiu of .\i.iscrai'uin airline liere^ulation see BTCE, i99oa. and more recency.- 
'orss-lh. I'JO.-ib. 



same general comments apply to the more disaggregated data of the second and third 
row of Table 5. It is however important to notice that the second group (Svdney Onlv) 
has significantly and consistently larger growth rates (except for load facto'rs) than the 
other group. This result tends to suggest that, since deregulation of the domestic airline 
mdustry, the more than proportional increase in traffic on the Svdnev routes has been 
accommodated by a more than proportional increase in flight frequency and a more than 
proportional increase in aircraft size. 

Table 6 and Table 7 display the evolution of aircraft movements and passengers at 
Sydney KS.A according to the different type.s of markets, respectively. This enables us to 
more accurately determine the factors that have driven the sustained growth in aircraft 
movements at Sydney KSA during the last decade. Impressive growth rates are achieved 
for each segment of the market in terms of both passengers and, although smaller, aircraft 
movements. Table 6 and Table 7 show that regional traffic has e.xperienced a phenomenal 
growth during the last decade at Sydney KS.A. In fact, Sydney KS.A has consolidated 
its position as the largest centre for regional traffic in Australia, with its share of total 
Australian regional trafllic increasing from 11.9% to 20.5% in terms of passengers, and 
from S.6% to 14.0% in terms of aircraft movements during the 19S9-1997 period. 'This 
result suggests that Sydney KSA attracted proportionally more regional traffic than 
other airports during the 19S9-1997 period. Deregulation has brought new regional 
airline operators (some of the largest entrants are in fact subsidiaries of incum°bents 
Qantas Airways and Ansett Airlines) and these operators have been clearly attracted by 
the larger catchment area of the Sydney basin, and by its ability to feed the domestic 
and international routes. 



Insert Table 6 and Table 7 here. 



In summary, the fundamental changes in competition policy (deregulation) have 
stimulated demand through a mix of lower fares and higher frequencv (see also. BTCE. 
1995a and Forsyth, 199Sb). Major Australian airlines have also increased the size of 
their fleet in order to meet this demand. Preliminary empirical evidence suggests that 
there has been little scope for major Australian airlines to reshape their nTtworks in 
order to gain economic efficiency (see also .Jatmika, 1999). Wichin this context. Svdnev 
KSA has been able to strengthen its position as che primary national and interna'uonal 
gateway, and to continue to experience impressive growth rates in aircraft movements 
and traffic, widiout however becoming a large US-style iuib airports Undoubtedly, this 
will have an impact on both airport capacity and aircraft noise, the primary concerns 
of this paper. In fact. Sydney KSA ha.s :h.e worsr, record in terms of the magnitude 
ot aircraft noi.-se on surrounding ooinnnmirirt'' arour.d maj^vr .Australian airports, T!ii> 
IS. ot course, not surprising given Sydney KSA':i g-ov.ti; ,u;d its |)roxiinity (location! to 
che center business district. Table S prox-i.ies an estimation of the population (private 

'Xevorr.liele^is. ihe.-e :,s -vidence of 5ome hiibbii;,' uLiv;-.;.- a: .r.-.ln.ry KS.\ {set also Seccioii o.:). 
"This ;= «^es~r;lj us::\i; ch.; A;:sr.va!iaii N.m<.- iZx:,o-i:r- in :-;■; ' '2' i ::',.?Aj .:]',, :uid 4u i 



clear tradeoff becween capacity e.\|)ansion and negative externalities. AUhou<^h the above 
figures suggest that, in this particular case, the cost of environmental externalities (based 
on present practise of quantifying them) is small compared to the projected economic 
benefits, transport decision making must be cognisant of principles of sustanability where 
economic, social and environmental factors and the mitigation of adverse impacts are 
included in the evaluation framework. 

4 Measures to Address the Externality Problem at 
Sydney KSA 

In order to address the externality problem at Sydney KS.A., an Environmental Impact 
Statement (EIS) was commissioned by the FAC, which subsequently satisfied Federal 
Commonwealth environmental legislation. Construction of the third runway was ap- 
proved subject to recommendations aimed at finding ways to reduce the unhealthy and 
socially disruptive impacts upon the residents and environment of Sydney. These recom- 
mendations have been detailed in the Draft Noise Management Plan and the Draft Air 
Quality Management Plan (Mitchell McCotter, 1994a, 1994b, 1994c, 1994d). Following 
the recommendation of the Draft Noise Management Plan, the .Australian Federal Gov- 
ernment adopted a Final Noi.se Management Plan (1996), not released to the public, 
which combines: (a) a list of measures to alleviate the noise problem in line with a tra- 
ditional direct regulatory approach ('command-and-contror) and, (b) the formulation of 
a noise levy on aircraft in order to raise the money for these measures. In contrast to 
the direct approach, the second type of instrument is more market-oriented. We discuss 
both approaches in turn. 

4.1 Direct Regulatory Approach ('Command-and-Contror) 

The 'command-and-control" approach involves the setting of technical and environmen- 
tal standards enforced via legislation without the aid of m.arket-based incentives. This 
has been so far the traditional and preferred approach adopted by airports and regula- 
tors when dealing with noise-related issues. For example, prior to the construction of 
the third runway at Sydney KS.A. the .Australian C^overnment implemented the gradual 
phasing-out of Chapter 2 aircraft to be completed in a seven year period from .Jan- 
uary 1995 to April 2002. In addition to this mandatory measure towards noise reduc- 
tion, the .Australian Government determined new measures specific to the Sydney KS.\ 
capacity expansion and its noise-related problem. The principal new resolutions chosen 
to be a part of the noise mitigation policy can be described as operational measures and 
administrative measures. The operational niensures inrhide: 



specific noise abateiiienc psocedurcs io- riirrmr. operalioiis and airport ground 
operations (e.g.. preferential runway use sy.5r.em.. pret'erentiai iiisht track use); op- 
erating restrictions and slot allocar,ions ^ i^s<enr.ial!y. a limit on bodi the number 
and type ot aircraft for domestic and international operations during curfev/ time 
; 1 iDm-Gamli; 



Following the recommendation of the Draft N'oise Management Plan (Mitchell Mc 
Cotter, 1994a 1994b). the Federal Commonwealth Government h^ adopted user char-^es 
following the Polluter Pays Principle on the aviation industry^ While the main aim ol a 
standard Pigouvian tax is economic efficiency (i.e., optimal levels of production and con- 
sumption), the mam objective of the Polluter Pays Principle, as formulated bv OECD 
m 19r2, IS equity: '"the polluter should bear the cost of measures to reduce pollution de- 
cided upon by public authorities to ensure that the environment is an acceptable state'' 
(quoted m VVallart, 1999). Since the empirical estimation of the environmental and 
financial impact due to airlines' operations is a far from exact procedure, most govern- 
ments, aviation and/or airi)ort authorities rely on an ad hoc formula to apply the Pollute^ 
Pays Principle^. In general, regulatory authorities follow the principles recommended bv 
International Civil .Aviation Organizacion (IC.AO) when netting environmental (mostfv 
noise- related) levies. IC.AO policy on environmental levies recommends that any en- 
vironmental levies on air transport which States may introduce should be in the form 
of charges rather than taxes, and that the funds collected should be exclusively applied 
towards mitigating the environmental impacts associated with air transport activity 
('•no fiscal aims behind the charges") (ICAO. 199Sa). More specifically, with respect to 
noise-related charges, ICAO recommends that the following principles should be applied 
(ICAO, Appendix A, 199Sb): 

• Noise-related charges should be levied only at airports experiencing noise prob- 
lems and should be designed to recover no more than the costs applied to their 
alleviation or prevention (charges should relate to costs). 

• Any noise-related charges should be a.s.sociated with the landing fee, possibly bv 
■ means of surcharges or rebates, and should take into account the noise certification 

provisions of Annex 16 (ICAO. 1993) in respect of aircraft noise levels. 

• Noise-related charges should be non-discriminatory between users and not be es- 
tablished at such levels as to be prohibitively high for the operation of certain 
aircraft. In addition, the charges should not discriminate against air transport 
compared with other modes of transporr,. 

Moreover, industry trade as.sociations like International Air Transport Association 
(LATA), Association ot European Airlines (AEA). and Airport Council International 

'^Lesislalion lo implemenl die noi.se char-e wcu? introduced earlv m 1995. and became effective I.,l- 
1, 1990 (.Aircraft .Noise Levy .\ct 199o). It is important to stres.? that under the .\ircraft Noise' Lev'.- 
Collection .Act 1995, Sydney KSA is the only -qualifyms airport' in .-Vustralia. Two conditions ar'e 
required to be a qualify, n; airport at a particular time: (a) at the t.me there is a public buildin- within 
a .o-unit contour, or a residence within a :lO-unit -oiitour shown on an .A.\EF previously prepar-d -or 
the area around the airport for a date after "hat Mine: and {b, r.ii. Commonwealth is funding at -hat. 
time, or has funded b.-tbre that -m,^, a no.s- ..meiioration pro^r.m for the airoort. Note that on- 
an airport has becom. a .ualify,,,. a,r.,o,T. u ,■..;,.,.:, . ,uaii:\;., airpor: ^ven' if it no longer meer. 
condition (ai (se- Art. (j,.>j 

^User char-es are usually ..-t lower than puiv l'-ouvian ta:-s. resulting in a hi-her level of exter- 
nality, all else equai. WMi'.n i :999) ,how. ^aat a .ui.optimai i;.e:- ;iKu-e level .:an result in the optimni 
pollution level, provided that its revenue is :,.,.d ;o:- abaurment ^;,-:-,iiu-, and that the use: ^har-e leve! 
is set adeqt;ately. 



199S, the LUR was set at Aii.sS IGo.lS. i.e. a nominal increase of G.6% throuo-hout fhf 
95-9S period" (see Talkie 11). 



Insert Table 1 1 here. 



Table 12 clearly shows that noisier aircraft pay more, all else being equal. The 
difference in the NLC between the B-737--200 and the B-737-400 is quite°striking. The 
levy for the B-737 Chapter 2 version is three times larger than the levy for the Chapter 3 
version. There are also important differences on a per passenger basis and, to a lesser 
extent, depending on the seating configuration of aircraft. According to Table 12, the 
difference between a Chapter 2 aircraft and a Chapter 3 aircraft in the charge per 
passenger is larger for smaller airplanes. Indeed, the charge pe/- passenger for a B-747 or 
a DC- 10 does not vary noticeably according to its Chapter certification. However, for 
aircraft in the range of 65-165 passengers, the charge /jer passenger is significantly larger 
for Chapter 2 aircraft. For example, for a similar capacity range, a F-2S has a charge 
of around AusS S.50 per passenger, in comparison to around AusS 2.00 for a BA-146. 
Given this result, one could argue that there is a strong economic incentive for airlines 
to phase-out smaller Chapter 2 aircraft first and/or to operate those aircraft in other 
city-pairs. We will come back to .some of these issues in Section 6. 



Insert Table 12 here. 



5.1 Properties and Advantages of the NLC Scheme 

According to OECD (1991) guidelines for the application of economic incentive instru- 
ments, there are a number of general criteria against which the various economic instru- 
ments can be normally evaluated. These criteria are: the environmental effectiveness 
principle; the equity principle; the (.Hatic and dynamic) economic efficiency principle: 
the administrative cost-effectiveness principle; and the acceptability principle. We exam- 
ine them critically with respects to the NLC at .Sydney KSA. In practice, these principles 
often conflict with each other, forcing the adoption of compromises and/or of innovative 
solutions. 

Table 12 for details on the noi.se metrics). The total number of NU corresponds to (10,000 • 6.44) - 
(90,000 • 1.96) = 240,800. If the total fi:u.Ls to bo ;,'enerate.:l for a particular year is .A.usS 40 million, 
the LUR (i.e. the S value of one iioi.se unit) would be eip-ial to .Au.s^i; Ififj.lO. .so that a Boeing 7.V-200 
would pay a SLC equal to .-\iis.'> IDTU.OU. wliil- a Boein- 7:37-l()(J .•in-crai"t -.vould p.ay a .N'Lc" equal r.o 
-Aus^i' .12(1.00 jKr landing. 

The .Aircraft .\oi.=e Levy An lyO-j ]uovid.> liiat. for ilie Hii.iucinl year ei'.din^ .June 1906. the LUR. 
should be less than .-Ui.sS 160.00. with a nia.\:iiiiai iiurr.M.se of lUVc, for the following year. 



the equUu pnncple seen,, deserved since t,he NLC does not confer a disproportionate 
burden on the lea.t well-ofF aircraft operators and/or aircraft users (passengers) (see 
Table 12). Whether the current NLC is (sufficiently) .fficrent in providing continuous 
incentives for noise nuisance reductio.is is rather difficult to answer <^iven inter alia 
financial, technological and operational constraints (see also Section 6 Ld Section 7). 

5.2 Potential Limitations of the NLC Scheme 

We see however several limitations with the current NLC scheme: 

1. The most significant issue is that the level of the noise levy is set bv the sum needed 
to fund compensation and not by the marginal cost that noise nuisances impose on 
society '. However, from an economic efficiency point of view, a noise levy should 
refiect the true marginal costs created by the externality, as well as the marginal 
abatement costs (which depend on the technology available for, e.g., engine hu°shk- 
its, windows insulation, etc.). Because the sum nee,U'^\ to fund compensation is set 
to vary each year (and eventually it is set to tend towards zero after a period of 10 
years), while the true marginal costs are likely to be more steady, the divergence 
from marginal cost pricing could be substantial in the medium/long run. 

2. The 265 (EPNdB) .^NL threshold level is arbitrary, and does not imply that only 
aircraft with an ANL greater than 265 induce noise environmental dama-es Noise 
levy exempt aircraft like the MD-90-:30 (with an ANL equivalent to 260), the Saab 
2000, the Fokker 50, and some versions of the BAel-16, are not exactly 'silent' 
aircraft, and therefore also induce negative externalities. Similarlv, the NLC does 
not apply to propeller aircraft or to helicopters. In designing the formula there 
was a strong desire to ^u-hieve a degree of comparability between the total funds 
raised from international and domestic/commuter operations" (FAC, p. 9. 6, 1990). 
Indeed, because domestic operations at Sydney KSA strongly outvJeigh interna- 
tional operations^^ and because domestic and regional operations use smaller air- 
craft than international operations, there wa.s a concern that the burden of the 
noise levy would proportionally be more important on domestic markets, unless 
small jet aircraft would be le.ss heavily taxed, or completelv exempt. Clearlv, 
from an economic effi.:ien<-y point of view it is fair that quieter aircraft should 
be taxed less. Whether it i.s desirable from an equir/.- point of view that laroer 
noisier aircraft are he.u'ily taxed (sur.-liarge) while r.hr smaller quieter aircraft are 
noise levy exempt (some sort of rebate) is debatable. One can argue that this 
scheme provides some incentive for aircraft substitution. However, we believe that 
this substitution is rather limited, because the more noise-efficient aircraft can 
have very different operational characteristics (i.e. size, range, etc.) than the less 
noise-efficient aircraft "\ 



■Jt, 13 thtii-efore not .-^ .iCaiulard Pi'^.jn'. i.tii i.ix. 
J-^IOT.OUO d0M,..t.c. :0.1,OOO vr^^^o^v.\ l\v-\u.. i„„s 2:.uV; -...--al :u,;,r,,on ;uul miiicarv fliglus, versus 
-l.j.OOO intrTi\.\rioii;il ii!Oveiii.:'ni..s for tlu' ii.s^-.u '.var 1 :;!ji'i- :'J!J7. 

'\\ v<il.M.A. althoii^i, iliin-rrnr, issue j. i.i;^ |„:...,h,i;:y -o ,:orr"cc ---.jriialiues through taxes on or 
iUDSkli-s -0 i-.-l;vC.-.| -:o>!> or i.rij.Jucr h^n ;.p ;■.■:•>-.•> ■ -^- ■ ^ Wiil.ai: .!-;■ ' [)<■ . 



7. Sydney KSA is, so Far, the (jiily aitpoi't in Australia to face tlie NLC. Because 
airlines operate on a spatial network, there is a need for cooperation amon<^ the 
different airports in the country, and maybe harnionization of the tax (at the 
national and sometimes international level). Otherwise, there is a potential for 
introducing discriminatory measure.s that distort competition and resource allo- 
cation. In fact, one potential operational effect of a locally-based NLC is for an 
airline to divert its noisiest aircraft to other route.s of its network where the noise 
restrictions are less stringent and/or where the financial penalty is more accom- 
modating. Note that from an economic point of view such an outcome could be 
acceptable if the external costs related to aircraft operations are lower elsewhere, 
a situation that could potentially arise in Australia (see Table S). 

6 Impacts on Aircraft Operators and on its Cus- 
tomers 

Major Australian airlines (Qantas and .Ansett) have upgraded and have expanded their 
fleets during the 90s, antl today their fleets, by and large, comply with the highest 
noise standards (although .Ansett still operated three (Chapter 2) F-2S in .June 199S, see 
Table 13). The mandatory phase-out of Chapter 2 aircraft coupled, to a lesser extent, 
with a higher NLC for Chapter 2 aircraft, has induced .Australian airlines to rapidly 
withdraw Chapter 2 aircraft from Sydney. Because the NLC /Jer passenger is significantly 
larger for smaller Chapter 2 aircraft, mainly F-2S and B-727, there has certainly been a 
stronger incentive to pluuie-uut these particular types of aircraft. However, we strongly 
believe that the main force driving the withdrawal of some Chapter 2 aircraft is the 
compliance with federal antl international laws and the completion of the aircraft life 
cycle, rather than the additional NLC. Industry sources suggest the effect has been the 
withdrawal on one aircraft type, namely the F:2S. 

Insert Figure 13 about here. 

For aircraft operators, the direct effect of the NLC is an increase in airport-related 
charges (part of the operating costs), and therefore a monetary transfer to the airport or 
the government authorities. Because both domestic and iat<-rnational .Australian airline 
markets are highly duopolistic, and demand for air transportation is fairly inelastic, air- 
lines are more likely to (ilirectly) pass a suljstantial fraction of the NLC on passengers. 
In fact, a noise charge of Aus$ 3.40 /n-/' passenger is automatically being imposed by in- 
dividual airlines at their discretion to i-ecover the costs they inciu' in paying the NLC at 
Sydney KS.A (F.AC. 199()). This charge npi)lies to every domestic/regional and interna- 
tional passenger landing at Sydney I\S.-\. With more than LQ million passengers inbouni.i 



new .scandard.; niusr l.t; acvisc;! .iii.j iiiiiili'ii'.iriit.r-d in oi>!r;r 'o c'.;ri) toial uoisf; levels. So far. ho■•'•■e■.■-;.^ 
[C.-VO aiiij 1I..S C'.iiiinii: Ll.-- oii .Aviarioii Eiu IrijiuuiMir.a! ri-otrt'-r.ion {(^-\EP), alr.hongii recognizinij :;•:"•.:. 
.le'.v noi.s.' c-.'rr.ilii;ai.iO!i srainlanls iim-.l l.) he- il.ni-iop'-i! 'liar prooi-i-ly take accouiu of cech!iolo'.;::a; 
pro,L;ri.'S>. aiv iinain:- ;.o ii^arli a .iiUmmimi.^ ■jii any >iii-!::;ic |,iropo.--al ro iiir.roihicr^ a new aoi.se suaii'.iar'i 
[ICAO. I'MSc). 



have that about 4S to 19:] international fliglits, and about '274 to 2,105 domestic/re'^ional 
flights might not be annually scheduled at Sydney KSA as a result of the NLC. Fn the 
total, around 322 to 2,298 flight movements could be annually diverted from Sydney 
KSA (between 0.132-0.943% of actual movements). Although these effects are small, and 
other factors lil<e exchange rates, avgas price, and domestic and international economic 
growth are more likely to influence the future trend of the air transport demand at 
Sydney KSA, our analysis indicates that depending on the price elasticities estimates, 
and on the amount of the NLC, the total number of aircraft movements may be curbed 
at Sydney KSA under a regime of NLC. 

Table 14 summarizes the lihely impacts on demand and on aircraft movements, had 
airlines impo.sed a per pa-s.senger noise charge of Aus-S C.SO or AusS 10.20 instead of 
AusS 3.40. Such an increase in the noise charge would have occurred had the actual 
LUR been set higher, as suggested by some local community advocates'-^ The results of 
Table 14 suggest that, when the price elasticity is valued at its high range, a reduction of 
around 3% of annual aircraft iiu)vements could arise under a per passenger noise charge 
of Aus!S 10.20. .AH in all, thi-se residt.s .sliow that the airline industry indirectly bears 
some of the social costs fi.-ssociar.ed with this niod(-' of transportation in the form of a loss 
of potential passenger revenues. .Similarly, s(;me airport-related charges are forgone for 
Sydney KS.A. 



Insert Table 14 here. 



7 Impacts and Distributional Consequences of Air- 
craft Noise 

7.1 Before the Long Term Operating Plan of March 1996 

Given aircraft types and aircraft noise characteristics, the allocation of aircraft to fii"-ht; 
paths-" ultimately determines the noise exposure of residents surrounding the airport. 
With the opening of the third runway [l(iL-34R] air traffic control had more flight path 
options (diversification argiunent) available with Sydney's airspace. However, the Labor 
Government in its determina.tion u\\ the tiiird runway EIS imi^osed an operational re- 
striction that there wcjuld he no takc-olfs r.o the north from i.lip new runway because o;' 
noise impacts on resitlents to the north. In early iUOo. the runwa\s available at Sydney 
KSA were as follows (see Figure 1): 

. Arrivals: 07-25. 16R-34L. and l6L-:^^R, 



-"[nd-ied, iliey ai-;iuecl dial a lii.:;;lier L[; R w.juUI li.ivr; raised additional revenue for die noise miiigar,;-: 
sclie.me, as well as providiii;^ a st.roii;.^er inceativif ro operate more iiotse-efficieui aircraft.. 
'Which is the res|)onsihiJity of air trnilic controllers of .AlrServices .-Vustralia. 



:^\ 



airport (MarrickviUe, Leichhardt. Ashfield. Druinnioyne, Hiiiiteis Hill, Lane Cove and 
Ryde). Parts of those suburbs to thf fast and west of the airport - Botany, Randwick 
Rockdale, Koijarah, and Hnistville - ol)tained a net gain in value of some AusS 200 
million from airci-aft noise leductions (Kiiihill Engineers, Table 2:}. 13, p. 23-31, 1990) 

In February 1995, soon after 0|)erations on the third runway brought home the redis- 
tribution of aircraft noise ami after consultation with community groups, the Australian 
Democrats (the third major political party in Australia) decided to push for a Senate 
Inquiry into Sydney's aircraft noise problems. The press release by New South Wales 
Senator Vicki Bourne (2S February 1995) said a public inquiry was essential given the 
"anger and distress"' caused by the opening of the third (parallel) runway. The Select 
Committee on Aircraft Noise in Sydney inquired, among other matters, into: the human 
impact of noise caused by aircraft movements following the ofjening of the third runway; 
reasons for discrepancies between the predicted and actual noise impacts (and proposals 
to prevent any such discrepancies occurring in the future); the likely effectiveness of the 
environmental management plans for Sydney KSA (and whether there are other poten- 
tially, effective measures, which could be implemented); and the potential for operations 
at the future Sydney West Airport (SWA) at Badgerys Creek to alleviate the impact of 
aircraft noise on "Sydney basin communities" (Commonwealth of Australia, p. 3, 1995). 

The Select (jommittee on Aircraft Noise in Sydney made comments critical of the 
environmental assessment of the third runway, and made recommendations on the op- 
erational measures implemented to reduce noise at Sydney KSA. One of the main rec- 
ommendations was the introduction of a legislative cap on annual movements at Sydney 
KS.A, and SO aircraft mo\-emeiUs /;f:r hoiu- is now Covernnient policy. 

7.2 Since the Long-Term Operating Plan of 1996 

On March 2, 1996, the Liberal/National Coalition won the Federal election (defeating 
the Labor Party) with a k'nidslide victory in the House of Representatives'^. The Sydne^j 
Morning Herald, on the day of the election, summed up what each party had to offer on 
Sydney KSA and on the future SWA at Badgerys Creek. The incoming government's 
election policy is reproduced in Table 17. 



Insert Table IT here. 



conducted for Sydney KSA. Tlir^ most, cu.wl >Mu;i,'.s (Abnljoa. 1077, ;uid the Draft EIS, L990) used i 
hedonic pi-icing mer.hod r.o ideiir.iiy -.he iinplicii. nru-.' aruudied to dilfeci'iit. variables by the house buyer. 
The Draft ELS .study (HJ9U) sampif-d :M4 hoii>.'s ii\ Boi.aiiy. .M.irrickvill,- and R.oc.kdale and compared 
prices in noise-aireci.^id are.-i.^ '.viri; .-.jiiipiu-aldL' nrn-cs diar. wi?rc not, aoi.se affected. The most recent 
study by .JLW R.rsearcii ami (■■onsi;ir.;iiicy P:y i.r.d (.Mitchell .McCotier. 109-11). .Appendix . I), sampie'l 
7.50 property tran.sactions in ['Ml .-.r.d i'.;!}-.' 11011;- r.!;e nortli-.south lli-iit path and compared price.s wit;-. 
near.by non-noLse arfected j)rop.-:T.i.-.>. .Vr-;:ar.v iir,-i:iliinis (Mopr.'ciation r.u.^s) in these latter two studi-s 
of th.e northc-rn suburb.s .ire siiniin.irized in Tablr' Hi. 

"In the Senate (i.'pper iloM.sej. r;..e Dr-iiiocrar.s, •■•.■itli S .seat.s continued to hold the balance of power 



West runway moiUlily comphiiur.s rlimbnl steieply to 6,5000 in July 1996 (Stao-e I of 
the LTOP). D('|)aftures !or tlie fifsr, tinir of nmway :J4R. (sw Flgme I) prompted S, 000 
complaints for the month of November 11)96. Complaints fell rapidly, and by November 
1997 about 2,500 monthly complaints were received. Implementation of Stao-e II of the 
LTOP in December 1997 pushed monthly complaints pa..st 9,000. Finally, the most re- 
cent data (October 1998) indicate that the number of monthly complaints stabilized at 
about 3,400 complaints (from 612 complainants). Table IS illustrates the distribution of 
aircraft movements according to the cardinal points: (1) before the construction of the 
third runway (Pre-Parallels, 199:]); (2) after the construction of the third runway under 
the Labour Government (Parallels, 1995); (:]) after the construction of the third runway 
under the Coalition Goverimient (.Stage I of the LTOP, end of 1996); (4) Stage II of the 
LTOP (Oct. 199S), and finally; (5) under the long-term Coalition Government's tar<^et 
for the LTOP. 



Insert Table IS here. 



Although it IS too early to provide a comprehensive economic, operational and en- 
vironmental as.sessment of the LTOP, the concept of noise .-iharing through a safe and 
efficient use of different oijerating mode.s of all runways is theoretically appealing. In 
fact, under specific circumstances public choice theory ha.s a strong argument in favour 
of 'externality distribution". .Assuming that the marginal e.xternal cost (MEC) associ- 
ated with aircraft noise is increasing^'' in the number of aircraft movements per runway 
(N), it is easy to show that total external costs are lower when aircraft movements are 
distributed over a larger number of runways. Let us consider the following graphical 
example. Figure 2 displays a linear (increasing) relationship between the number of 
aircraft movements /jer runway and the marginal external cost, i.e., MEC=a+b-N, with 
a>0, and-" b>0. First, a.ssume that all the runways have identical characteristics, and 
that a total ot 120 aircraft [noven)ents [pfr time unit) are e<pially shared among two 
runways (a situation which would depict the 'Parallels" regime in 1995). Since total 
external costs correspond to twice the area under the ME(; of Figure 2, it is ea^y to 
show that total external costs correspond to .3,600-b. Now, let us assume that the same 
total ot 120 movements are equally shared among the three runways (a situation which 
would rather represent the LTOP's target). A straightforward computation shows that 
total external costs are reduced to 2,40U-b, all else equal. In other words, when the 
marginal external cost prr runway is rising, sharing the traffic over more runways can be 

a .sufvey of" :!,.o7o .-\ii.su;\li.-ui r.--;iil.'ius (llr'dn :md Riilli!n. hJS'J). T^ai- oT aircraft cra-shini;, personality, 
socio-rtcoiiormc sint.ii.s aiui imi-Im foru> .ire l'.icl.or> thai iii(liir-ii<;.' uliri I.im- aiiiioyanc;- from aircraft is 
reported. 

"' Tlui IS .-,1 .stiuul.u-d a.i^iiiuii[;oii i:i i-ir.n-oiiiiici;Ml rconoiiiK'.--, an i ; i, :s nio.st lik.dv to appiv in li;-^ 
case at liaiid, altliijugli it i.s iil; imaf.-ly an rrmpincal iiiif-riMuii- T!ie icini.-itu; iur"e in tlie number o: 
noise compiaiiu.s alter the oiK-iiiii;.; ..i;' i.iie third riiiiu-ay crt.ninly ^iipporr.> riie .standard assumption of 
increasing mar;;iaai e.xteriial co-r.s. 

For simplicity'^ sake, and ■.virhout loss of .:j;i.'iieialit\ . we ^aii set a=0. as m Fi'mre 2. 



from the noise levy, if then^ is a cle.siiK to fully apply tlie Polluter Pays Principle^^. 

In an ideal world, one would aim at coin[)reliensively addressing (i.e. internalizing) 
the full environmental cost.s a.s well a.s benefits a.ssociated with aircraft operations. Such 
an economic 'First best' would be rather difficult to achieve given the complexity of the 
problem at hand (uncertainties, multiple constraints, etc.). The Polluter Pays Principle 
applied in Sydney KS.A, can be described a.s a "n-th best" given the different constraints. 
To the best of our knowledge, whether .■sufficient static and dynamic eflBciency (i.e., 
incentive to reduce aircraft noise e.x'ternalities) is achieved under the current scheme is 
difficult to assess. Similarly, even if efficiency is achieved under the current scheme, 
local/regional economic uptimality does not necessary imply global optimality. The 
fact that the scheme chosc-n at Sydney KS.A cotubines 'user charges' (i.e.. Polluter Pays 
Principle) and some degree of 'internalization" seems to us appealing in the case at hand 
(i.e. in the very contentious and comple.\- context of an airport). 

The main contribution and originality of this paper is the integration of the various 
aspects and dynamics driving the economics of air transport in relation to airport infras- 
tructure and operations and the associated environmental externalities. For a number of 
reasons (e.g. availability of data, transparency, originality of the scheme, etc.) Sydney 
KSA provides a unique framework for analyzing this complex issue. On the other hand, 
by focusing on the specifics of Sydney KSA, we . e able to discu.ss broader principles 
and policy issue.s likely to be relevant for other maj r airport.s around the world. 

9 References 

ABELSON, P.W., 1977, The Impact of Envirnnmenial Facton on Relative House Prices, 
Bureau of Transport Economics, CaiilM?rra. 

AIRSERVICES AUSTRALIA, 1!)D7. 

BTCE, 1992, Derecjulation of Dointvlic Afintion in Australia 1000-1905, Information Sheet 
6, Bureau of Transport and C^oniinuuicatioiis Ecouomics. CanlKM-ra. 

BTCE, 1994, Adequacy of Transport Infrastniclure: Aiqiorts, Working Paper 14.4, Bureau 
of Transport and Connnunicatioiis Economics, Canberra. 

BTCE, l99oa. Quality of Service in Australian Passenrjcr Aviation. Report, SO. Bureau of 
Transport and Communications Economics, Canberra. 

BTCE, 199ol), Demand Elastu.-itii:.^ for Air Travel to and from Australia. Working Paper 20. 
Bureau of TransiJorr. aiicl Communicntious Economics. Canberra. 

COMMONWEALTH of .AUSTRALIA, !09.-). Falluuj on D,af Ears? Report of the Sde'-.l 
Comniitttc on Aircraft .Wji.-^i. in S'jdm ;j. ("aiiberi-a: .Seiiai.e Piniicing Unit, Parliam-?r.r 
Kou.se. 



35 I 



Mt. is iarrti-p.-suii:r co not..' r.liai. r.hf principlir of an i^ini.-i.iion-rrjlat.r-d ■■h.-vi-^e. ai!.hou;;h noi y,;'. fui'iy 
enclorseil by ICAO. lin.s n?ci.'nily ;.iaiiieil .some f.-ivonr.ibli' '^rouiui ainoii'j, rhe im.ernacioiial aviacio.n corr,- 
munity (IC'.AO, llJOcSc). 



ICAO 199;3, InterncUior^al St.ndarcb and Recommended Practice., Environmental Pmt 

tu>n, Annex 16 to the Convention on International Civil Aviation, Internat o al Ch 1 
Aviation Organization, Montreal. '"-ernational Civil 

ICAO 1998a, Agenda Item 32: Taxation of International Air Transport, Report by the Coun 

A32-WP/01, September, International Civil Aviation Organization, Montreal. 

ICAO 1998b, Aoenda Item 21: Environmental Protection, Report on Environmental Charges 
and T^es Assembly - 32nd Session, Working Paper A32-WP/35, September Internl 
tional Civil Aviation Organization, Montreal. 

ICAO, _ 1998c Arjenda Item 21: Environmental Protcctron, Report of the Executive Com- 
mittee, Assembly - 32,k1 Session, October, International Civil Aviation Organization 
iVlontreal. 

ITS, 1997, Report on the Economic Sirjnificance of Sydney International Airport, Institute of 
iransport btudies, The University of Sydney, Australia. 

JATMIKA, H.E., 1999, Major Ai,-port Operations and Land Use Interactions: A Spatial 
Development Evaluation with the Inclusion of Economic and Environmental Aspects 
mimeo, University of New South Wales, Sydney, Australia. 

KINHILL ENGINEERS, 1990, Proposed Third Runway (Kinrjsford Smith) Airport - Draft 
Environmental Impact Statement, Sydney: Sydney KinhiU Engineers for the Federal 
Airports Corporation. 

KINHILL ENGINEERS, 1991, Proposed Third Run.ay (lungsford SnnthJ Airport - Draft 
Environmental Impact Statement Supplement, Sydney: Sydney KinhiU Engineers for the 
l-ederal Airports Corporation. 

LEVESQUE, T J., 1994, 'Modelling the Effects of Airport Noise an Residential Housing 
Markets, Journal of Transport Economics and Policy, vol 2S(2), p. 199-210. " 

MITCHELL McCOTTER, 1994a, Sydney (Kinysford Smith) Avport: Draft Noise Man- 
agement Plan Volume 1 - Summary Report, Sydney: Mitchell McCotter Consultant to 
the steering Committee. 

MITCHELL McCOTTER, 1994b, Sydney (Kinysford Smith) Airport: Draft Noise Man- 
agement Plan Volume 2 - Technical Report, Sydney: Mitchell McCotter Consultant ro 
the bteering Committee. 

MITCHELL McCOTTER, 1994c, Sydney (Kinysford Smith) Airport: Draft Air OualU 
Management Plan. Volume 1 - Summary Report. Sydney: Mitchell McCotter 1-' Asso 
ciates and Peter W . Srephen.so,, ,t Associates. Consultants to the Steering Committee. 

MITCHELL McCOTTER. 10')4d. Sydnuj (Kiny.ford Smith l Au-uort: Draft .Air Onr.ia 
.Management Plan, \olnnu: J - TccUn^m! Rcuorr. ^yn^^.y: \\]r.ciw\l McCotter i'Aiso- 
ciates and Petor \V. Sr.'plieiison ,v: As.ociac^s, Coiisuitnnu ro ■!,<. Steering Comi;-;-:--. 

NERO, G., and BLACK, J. A., vm. iiub-aud-Spok,' v.nvvork., r.,i,l r.ue Inclusion of Zv.- 
vironmeutai Costs on Airport Pricing." Transporuuion Research Part D. Transport <:r.a 
Environment. vo!.:ji'.)), pp.275--29G. 



y 



7 



10 Appendix 

Table 1: Recent Key Dates and Events for the Ausf.i-alian Airline Indus t r y 
Date Events 



1952 Introduction of tlie Two-Airliae Policy: This is a legislation to limit 'uneconomic' (i.e., destructive) 

competition between tlie two domeiitic airlines, the publicly owned Trans Australia Airlines (TAA) 
and the private Australian National Airways (ANA). 

1957 Private Ansett (then a regional airline) takes over ANA and forii\s a new airline Ansett-ANA, 

which was later renamed Ansett Airlines. 

Late 19S9 Pilots' dispute affecting the Australian domestic market. Traffic decline by nppro.'C. 20% in 1990. 

Nov. 1990 End of the Two-Airline Policy. The .Australian domestic market is completely deregulated. 

Feb. 199'.' iVterger between publicly owned Qantaa and Australian .Airlines ( AA) (e.\-TAA renamed in 1986). 
The new Qantas becomes the only m.ijor operator at that time with both a domestic and an 
international network. 

Government adopts the principle of nuiltiple designation in international air services agreements. 
This enables Ai\sett Australia to lunch its international operations in .Sept. 1993. while Qantas 
looses its status as sole designated Hag-cairier. 

Mar. 1993 Government sells to British Airways a 25% st.-\ke in Qantas. Subsequently, both airlines form a 
strategic alliance. 

Nov. 1994 Inauguration of the third runway at Sydney KSA. 

Jan. 1995 Mandatory phase-ovit of Ch,-xpter 2 aircraft to be completed by April 2002. 

Jul. 1995 Government sells to the public its remaining 75% of equity in Qantas. thereby becoming fully 
privatized. 

Oct. 96 Private Air New Zealand piucliases a 50% stake in Ansett Australia from TNT Coi-poration. 

.Nov. 96 Creation of a Single (Trans-Tasinan) Aviation iVIarket between Australia and New Zealand. 

Jul. 1997 Government sells 'Phase I' airports (Brisbane, Melbourne, and Perth) for long tei-m lease. 

Jun. 1998 Government sells 15 'Phase IT .lirpoits (Adelaide, Canberra. Coolangatt.i. and Hobart are among 
the largest airports) for long term lease. 



So'irrc: Various including,inl«r alia. Fiiidlay (1996), Hooper and Findlay (1997) and. Forsyth (l99Sa). 



-'fi 



Table 4: Top 20 Australian Aidine City-Pai. Markets foL- Fiscal Years 1991 and 199S 
(Ending on June 30) ^-^-^^ 



City-paiF 



MclbourbcSydney ,0. ■.3.355 J-M-.O -JJS/jai .,.S0«.385 3,913,527 6 632 701 

Bnsban.-M. bourn. 1,3S1 6,G01 ,5,72S 630,15-. 1,657,0,5, ^13 7." ' 060 396 

Adea.cIe-M.lbom-ue 6.,3 9.049 13,752 S90.66S l.300.1>9 l.-2U19^ 'xfslsll 

CooUngaUa-S.cL.ey 6S0 S^OO U.379 710,727 1.275.SS1 0..2^ ' '! 

Bn.banc.C.an. 1,391 4,677 9,037 408,567 958.401 552 592 'Sos 

Perth-Sydney 3,2S4 3 iSt 7 irj T»e .-n- .^,..,- i.JaJ.U.a 

.j.—jt .j,io-^ i,lt>4 3j5,b9o 916.027 491 7sq i iqtoc 

Melbou...PercU 2,706 4,771 6,914 516.096 S49:22S 704;U6 ! ' 

Caaberra.Syd.ey 236 9£27 22,168 5G2.651 8.30.076 953,665 1,322'573 

Hobart-Me bourne STi TJTr ojm TtUTt 7:18:^9 660:1530 iloTST 

Canberra-Melboume 470 7,108 9,486 443,271 695,580 743,748 1,080093 

Ca.rns.Syd..ey 1.97, .,.,35 4„S.57 131,495 644,.S8, ,66,951 940060 

Melbourne-Coolangatca 1,33U 2.728 4,408 258,497 543,722 335.026 699'644 

Br.sbane.Tow,.sv,lle M12 4,352 4.76U 337.829 448.111 ^33.465 .589620 

Uuce,ton.Me boui^;^ ??^ ^r^, ^35 52l7^Ji iTITil^l J=^;:^, j:^^^=^ 

Adela,de.PcrtU o.^o 2,723 4.3ir> 248.777 405,376 .339,490 555 234 

Bnsbane-RockJ.ampton SlS 2.907 5,593 176,915 243,644 273,468 40 053 

Bnsbane-MacUy 797 1.713 3,983 91.287 223,019 142.407 332;27-' 

Kalgoo.l.e.P.rtl. 538 1,4.53 4,030 89.490 190,527 106 107 31S 607 



JJiii! 129,107 210.061 11,750.474 21,307,5 



■S.S-l 16.605.892 29,200,501 



S^'ircr : BTCE, 1995a, and DTRD. 199Sa. 



Table 5: Summary Scati.stics for Top 20 City-Paii's 1991-9S, VVeiohted Average by Grouo 
of City-Pairs 



UlCy-pail- Ul-OUp -SLIkui) .VUu 9,-98 P.v^- 91-98 .S«.-.Ls 91-98 LF LF AS XT 

% Growcli % CrOH-th % Gi-o«cli ,99, ,993 ,991 1993 



,998 



T0C..IT0P ,.039 +75.570 +93.2% +91.8% 70.8% 73.0% 135 152 

Sydney Only (,) 962 +79.1% +98.3% +96.0% 70.2% 72.6% 147 167 

OcUers Except Sydney (131 1.1 55 +71.1% +S3.7% +85.6% 71.5% 73.6% 117 ,29 

S^jirci: BTCE. 1095a. and DTRD, 1998a. 

-Vof^.- SL=Sta5e Len^iU, Mvts=,Moven,enLs. Pa.\- = Pa:isensers. LF=Lo,vl F,-v,:t.>,-. AS = Aircraft Size. 



Table S: Approximate Pop,ilar,ion.s (PrivaLe Dwellings) Exposed to Aircraft No 
Measured by Australian iN'oise Exposure Index [ANEI], 1990/91 



ise as 



Citv 



ANEI 
•20-25 2.5-30 30+ 



ijytlney- 4oM0 io.OOO 9,000 
Adelaide 14,500 7,400 4,100 
Melbourne 14.900 IJOO 300 



S£nrcc: Federal Aii-porls Corpoi-atiou. personal co.u.uuulcalion; AirServices Australia. 1997. p.ioO. 
/Voit; -In 1995 the conesponding numbers for Sydney are: 68,400; iO.MO; and 1 1,000. 

Table 9: Annual Practical Capacity versus Actual Movements (includin<r General Avia 
tion) at Sydney KSA (100s) ° 



1990 1991 199> 1993 1994 



Practical Capacity 26so 2i>so isso jcso 26so 
Actual Movements isso isoo 2030 2220 2270 



1996 199G 1997 1998 2003 



3530 3530 3530 3530 3530 
2420 2560 2640 2640 3300 



Sour^: Various, including DTRD. 1993... BTCE, 1994. ,V(itchell Cotter, Table 2.1, p.2.9. 1994b. Projections for the 
year 2003 from the LTOP (1996) assuming 4.1% increase per year. 



Table 10: Evolution of Revenue Collected from Aircraft Operators and Expenditure for 
Noise Mitigation Scheme (in million of AusS). Fiscal Year Endin- in .Ji 



lune. 



E.vpenditure Revenue 

1994-95 AusS' 24.2 — 

1995-9C Aus$ 62.3 Aus$ 22.1 

1996-97 AusS 49.0 Aus^' 38 7 

1997-9S Aus§ 6S.4 AusS 39.4 



ionrc,.- Personal co.nmu.ucation with senior olhcer (Harry Carroll) at AirService.s Australia and DTRD, 1998b. 



:;o 



Table 13: Major Australian Airline Fleets (Including Regional Subsidiaries), June 98 



Aircraft Type 


Qantas 


Ansett 


B-747- 






31 




3 


B-7C7- 






26 




13 


B-737- 






3S 




'■)■•> 


A-300- 






4 




— 


A-320- 






- 




19 


BAel46- 






14 




11 


F-2S' 






_ 




3 


DHC-S- 






16 




- 


DHC-G- 






■D 




- 


BA-JS31 






4 




- 


Slioi-ts SD360 






I 




- 


Cessna 0404 T 


lean 









- 


lotal 






147 




71 


Sonrce: Aii'l 


lies 


annual 


reports 


199S 





Notet : 'All types included. 



Table 14: Effects on Demand and on Aircraft Movements from Different Noise Char°-es 



Type of 
market 



Reduction in demand 
from a per pax charge of: 



AusS 3.40 



.^usS (>.S0 



AusS 10.20 



',500-30,300 1-5,210-60,770 22,730-91,120 



International 

Domestic 

+ Regional 28,500-219,000 57,120-137,920 s5,gso-g5g,sso 



Total 



36 ,000- 2 -lO ,300 72,330-498,690 108,460-748,000 



Reduction in aircraft movement 
from a per pax charge of: 



Aus.'!;3.40 AusS 6.30 



AusS 10.20 



48-193 97-3S7 

274-2,105 549-4,211 



145-580 
824- 6,316 



322-2.298 596-4,598 



969-6,896 



Note: Calculations are based on price ela.stii;ities r.-inging from 0.5-2.0 and from 0.3-2.3. ,ind on a 'representative' 
round-trip air fare of .A.us.$ 1500 and Aus.S oOO; for international and <lomestii: markets respectively. 



Table 15: Number of Occupied Private Dwelling Types in the 20 ANEF and Above Con- 
tours for the Base Case (19SS) and the Long Term (2010) Parallel Runway Operations 

Location Relative 19SS 2010 % Change 
to ■A.irport 



North 
South 
East 
West 



iot<- 



23.15S 33.39S ^44.2 

1.071 1,236 ^1.5.4 

23.3S4 1,44.5 -93. S 

24.326 1,6S3 -93.1 



71.939 37.(62 



-47.0 



5Tircc ; BAied on Kiiiliill Engineers. Table 23.9, i).2.'5-22. 1090. 



32 



Figure I: Sydney KSA Runways System 




■34